diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,131530 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.99997716868914,
+  "global_step": 218997,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 19.908,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 20.1929,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 19.4791,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 19.2878,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 18.7751,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 18.2198,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 17.0716,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 16.3735,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 15.0237,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 13.7339,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 12.5948,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 11.3323,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 10.3432,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 9.6996,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 9.2018,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 8.7387,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 8.1843,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 7.5949,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 7.0324,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 6.5224,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 5.9935,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 5.5401,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 5.2945,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 4.949,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 4.821,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.6842,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.591,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.5478,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 4.4621,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 4.4744,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 4.4713,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.4719,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3e-05,
+      "loss": 4.3722,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.6032,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 4.4406,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6e-05,
+      "loss": 4.4667,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-05,
+      "loss": 4.3426,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8e-05,
+      "loss": 4.3965,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 4.3686,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 4.3819,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-05,
+      "loss": 4.3469,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2e-05,
+      "loss": 4.3828,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-05,
+      "loss": 4.3914,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 4.4878,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 4.3684,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.3271,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 4.4009,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 4.3349,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 4.3091,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 4.3317,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999238350572e-05,
+      "loss": 4.317,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998476701144e-05,
+      "loss": 4.3356,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997715051716e-05,
+      "loss": 4.2739,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996953402288e-05,
+      "loss": 4.3315,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9996191752859997e-05,
+      "loss": 4.2693,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9995430103431994e-05,
+      "loss": 4.1856,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9994668454004e-05,
+      "loss": 4.2378,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999390680457599e-05,
+      "loss": 4.2703,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999314515514799e-05,
+      "loss": 4.2246,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999238350571999e-05,
+      "loss": 4.1694,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999162185629199e-05,
+      "loss": 3.9582,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9990860206863986e-05,
+      "loss": 4.1133,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999009855743599e-05,
+      "loss": 4.1082,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998933690800798e-05,
+      "loss": 4.138,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988575258579985e-05,
+      "loss": 4.0601,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998781360915198e-05,
+      "loss": 4.1024,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998705195972398e-05,
+      "loss": 3.8193,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998629031029598e-05,
+      "loss": 3.9659,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998552866086798e-05,
+      "loss": 3.9218,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998476701143998e-05,
+      "loss": 3.9093,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9984005362011976e-05,
+      "loss": 3.7757,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9983243712583974e-05,
+      "loss": 3.8079,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998248206315597e-05,
+      "loss": 3.705,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981720413727975e-05,
+      "loss": 3.6058,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980958764299966e-05,
+      "loss": 3.6433,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998019711487197e-05,
+      "loss": 3.6646,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997943546544397e-05,
+      "loss": 3.576,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9978673816015965e-05,
+      "loss": 3.5351,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997791216658796e-05,
+      "loss": 3.3862,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997715051715997e-05,
+      "loss": 3.4665,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997638886773196e-05,
+      "loss": 3.3142,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997562721830396e-05,
+      "loss": 3.3298,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997486556887596e-05,
+      "loss": 3.193,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997410391944796e-05,
+      "loss": 3.2315,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9973342270019954e-05,
+      "loss": 3.1504,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997258062059196e-05,
+      "loss": 3.145,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9971818971163956e-05,
+      "loss": 3.1454,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9971057321735954e-05,
+      "loss": 3.1044,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997029567230795e-05,
+      "loss": 3.0806,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996953402287995e-05,
+      "loss": 3.0912,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996877237345195e-05,
+      "loss": 3.0143,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996801072402395e-05,
+      "loss": 2.8725,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996724907459595e-05,
+      "loss": 2.9242,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9966487425167945e-05,
+      "loss": 3.0031,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996572577573995e-05,
+      "loss": 2.9484,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996496412631194e-05,
+      "loss": 2.794,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9964202476883944e-05,
+      "loss": 2.8002,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996344082745594e-05,
+      "loss": 2.8263,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996267917802794e-05,
+      "loss": 2.7243,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996191752859994e-05,
+      "loss": 2.7113,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996115587917194e-05,
+      "loss": 2.91,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996039422974393e-05,
+      "loss": 2.6905,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9959632580315936e-05,
+      "loss": 2.7475,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995887093088793e-05,
+      "loss": 2.6774,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995810928145993e-05,
+      "loss": 2.8196,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995734763203193e-05,
+      "loss": 2.8821,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995658598260393e-05,
+      "loss": 2.617,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995582433317593e-05,
+      "loss": 2.696,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995506268374793e-05,
+      "loss": 2.5701,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9954301034319925e-05,
+      "loss": 2.5842,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995353938489192e-05,
+      "loss": 2.7802,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995277773546393e-05,
+      "loss": 2.6599,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9952016086035924e-05,
+      "loss": 2.6288,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995125443660792e-05,
+      "loss": 2.5988,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995049278717992e-05,
+      "loss": 2.581,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994973113775192e-05,
+      "loss": 2.6425,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9948969488323914e-05,
+      "loss": 2.4936,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994820783889592e-05,
+      "loss": 2.548,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9947446189467916e-05,
+      "loss": 2.6893,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994668454003991e-05,
+      "loss": 2.5482,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994592289061191e-05,
+      "loss": 2.448,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994516124118391e-05,
+      "loss": 2.6024,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9944399591755906e-05,
+      "loss": 2.4648,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994363794232791e-05,
+      "loss": 2.5336,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994287629289991e-05,
+      "loss": 2.4472,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9942114643471905e-05,
+      "loss": 2.4923,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99413529940439e-05,
+      "loss": 2.3767,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99405913446159e-05,
+      "loss": 2.697,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9939829695187904e-05,
+      "loss": 2.4795,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99390680457599e-05,
+      "loss": 2.4553,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99383063963319e-05,
+      "loss": 2.4628,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9937544746903896e-05,
+      "loss": 2.5527,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99367830974759e-05,
+      "loss": 2.5529,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993602144804789e-05,
+      "loss": 2.3136,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9935259798619895e-05,
+      "loss": 2.4799,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993449814919189e-05,
+      "loss": 2.4711,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993373649976389e-05,
+      "loss": 2.4319,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993297485033589e-05,
+      "loss": 2.4269,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993221320090789e-05,
+      "loss": 2.463,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993145155147988e-05,
+      "loss": 2.367,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.993068990205189e-05,
+      "loss": 2.3382,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9929928252623885e-05,
+      "loss": 2.4309,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992916660319588e-05,
+      "loss": 2.3107,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992840495376788e-05,
+      "loss": 2.4676,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9927643304339884e-05,
+      "loss": 2.4222,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992688165491188e-05,
+      "loss": 2.3188,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992612000548388e-05,
+      "loss": 2.2931,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9925358356055876e-05,
+      "loss": 2.4418,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9924596706627874e-05,
+      "loss": 2.395,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992383505719988e-05,
+      "loss": 2.4055,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9923073407771875e-05,
+      "loss": 2.3604,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992231175834387e-05,
+      "loss": 2.3424,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992155010891587e-05,
+      "loss": 2.3389,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9920788459487874e-05,
+      "loss": 2.3181,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9920026810059865e-05,
+      "loss": 2.3613,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.991926516063187e-05,
+      "loss": 2.2479,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.991850351120387e-05,
+      "loss": 2.4203,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9917741861775864e-05,
+      "loss": 2.3793,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.991698021234786e-05,
+      "loss": 2.3484,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9916218562919866e-05,
+      "loss": 2.1378,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.991545691349186e-05,
+      "loss": 2.2506,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.991469526406386e-05,
+      "loss": 2.2772,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.991393361463586e-05,
+      "loss": 2.2877,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9913171965207856e-05,
+      "loss": 2.3131,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991241031577985e-05,
+      "loss": 2.194,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991164866635185e-05,
+      "loss": 2.1883,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9910887016923855e-05,
+      "loss": 2.131,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991012536749585e-05,
+      "loss": 2.0833,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990936371806785e-05,
+      "loss": 2.1511,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990860206863985e-05,
+      "loss": 2.1787,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990784041921185e-05,
+      "loss": 2.1199,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990707876978384e-05,
+      "loss": 2.1944,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990631712035585e-05,
+      "loss": 2.1185,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9905555470927844e-05,
+      "loss": 2.1862,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990479382149984e-05,
+      "loss": 2.2729,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990403217207184e-05,
+      "loss": 2.2283,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990327052264384e-05,
+      "loss": 2.1885,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9902508873215834e-05,
+      "loss": 2.0927,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990174722378784e-05,
+      "loss": 2.1225,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9900985574359836e-05,
+      "loss": 2.0893,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990022392493183e-05,
+      "loss": 2.0164,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.989946227550383e-05,
+      "loss": 2.1532,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9898700626075835e-05,
+      "loss": 2.0983,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9897938976647826e-05,
+      "loss": 2.3654,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.989717732721983e-05,
+      "loss": 2.1112,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.989641567779183e-05,
+      "loss": 2.0267,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9895654028363825e-05,
+      "loss": 2.1427,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989489237893583e-05,
+      "loss": 2.1387,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9894130729507826e-05,
+      "loss": 2.1847,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9893369080079824e-05,
+      "loss": 2.066,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989260743065182e-05,
+      "loss": 2.1145,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9891845781223826e-05,
+      "loss": 2.1275,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9891084131795816e-05,
+      "loss": 2.151,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989032248236782e-05,
+      "loss": 2.0696,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988956083293982e-05,
+      "loss": 2.0398,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9888799183511816e-05,
+      "loss": 1.9605,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988803753408381e-05,
+      "loss": 1.9539,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988727588465582e-05,
+      "loss": 2.0076,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988651423522781e-05,
+      "loss": 2.0771,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988575258579981e-05,
+      "loss": 2.1204,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988499093637181e-05,
+      "loss": 2.0624,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988422928694381e-05,
+      "loss": 2.1589,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9883467637515805e-05,
+      "loss": 1.9711,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988270598808781e-05,
+      "loss": 1.9487,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9881944338659806e-05,
+      "loss": 2.061,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9881182689231804e-05,
+      "loss": 1.9713,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.98804210398038e-05,
+      "loss": 2.1053,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.98796593903758e-05,
+      "loss": 1.9711,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98788977409478e-05,
+      "loss": 2.101,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98781360915198e-05,
+      "loss": 2.23,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98773744420918e-05,
+      "loss": 2.077,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9876612792663795e-05,
+      "loss": 2.1825,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987585114323579e-05,
+      "loss": 1.9123,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987508949380779e-05,
+      "loss": 1.9667,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9874327844379794e-05,
+      "loss": 2.1362,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9873566194951785e-05,
+      "loss": 1.9384,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987280454552379e-05,
+      "loss": 2.0334,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987204289609579e-05,
+      "loss": 2.0166,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9871281246667784e-05,
+      "loss": 1.974,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987051959723978e-05,
+      "loss": 1.9085,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9869757947811786e-05,
+      "loss": 2.0255,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986899629838378e-05,
+      "loss": 1.9908,
+      "step": 2220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986823464895578e-05,
+      "loss": 1.9622,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986747299952778e-05,
+      "loss": 1.9998,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9866711350099776e-05,
+      "loss": 1.8649,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986594970067178e-05,
+      "loss": 1.9896,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986518805124378e-05,
+      "loss": 1.9784,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9864426401815775e-05,
+      "loss": 1.7652,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986366475238777e-05,
+      "loss": 1.9242,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.986290310295978e-05,
+      "loss": 1.8951,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.986214145353177e-05,
+      "loss": 2.0146,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.986137980410377e-05,
+      "loss": 1.9712,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.986061815467577e-05,
+      "loss": 2.1195,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985985650524777e-05,
+      "loss": 2.0771,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9859094855819764e-05,
+      "loss": 2.1057,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985833320639177e-05,
+      "loss": 2.1043,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985757155696376e-05,
+      "loss": 1.8596,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985680990753576e-05,
+      "loss": 1.7523,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985604825810776e-05,
+      "loss": 1.8939,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985528660867976e-05,
+      "loss": 1.8492,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9854524959251756e-05,
+      "loss": 2.1039,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985376330982376e-05,
+      "loss": 1.7899,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985300166039575e-05,
+      "loss": 1.7348,
+      "step": 2430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9852240010967755e-05,
+      "loss": 1.885,
+      "step": 2440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985147836153975e-05,
+      "loss": 2.0792,
+      "step": 2450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.985071671211175e-05,
+      "loss": 1.8665,
+      "step": 2460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9849955062683754e-05,
+      "loss": 2.0417,
+      "step": 2470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984919341325575e-05,
+      "loss": 1.9013,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984843176382775e-05,
+      "loss": 1.9101,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9847670114399746e-05,
+      "loss": 1.8856,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984690846497175e-05,
+      "loss": 2.0116,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984614681554374e-05,
+      "loss": 1.9585,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9845385166115746e-05,
+      "loss": 2.0424,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984462351668774e-05,
+      "loss": 1.7606,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984386186725974e-05,
+      "loss": 1.7255,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984310021783174e-05,
+      "loss": 1.9481,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984233856840374e-05,
+      "loss": 1.9222,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984157691897573e-05,
+      "loss": 1.8009,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984081526954774e-05,
+      "loss": 1.8186,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.984005362011973e-05,
+      "loss": 1.9381,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983929197069173e-05,
+      "loss": 1.7261,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983853032126373e-05,
+      "loss": 1.7136,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983776867183573e-05,
+      "loss": 1.8508,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9837007022407725e-05,
+      "loss": 1.8126,
+      "step": 2640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983624537297973e-05,
+      "loss": 1.8503,
+      "step": 2650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9835483723551726e-05,
+      "loss": 1.8932,
+      "step": 2660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9834722074123724e-05,
+      "loss": 1.807,
+      "step": 2670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983396042469573e-05,
+      "loss": 1.9108,
+      "step": 2680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983319877526772e-05,
+      "loss": 1.721,
+      "step": 2690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983243712583972e-05,
+      "loss": 1.8462,
+      "step": 2700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983167547641172e-05,
+      "loss": 1.7539,
+      "step": 2710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.983091382698372e-05,
+      "loss": 1.8146,
+      "step": 2720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9830152177555715e-05,
+      "loss": 1.7702,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982939052812772e-05,
+      "loss": 1.9241,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982862887869971e-05,
+      "loss": 1.9108,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9827867229271714e-05,
+      "loss": 1.8127,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982710557984371e-05,
+      "loss": 1.8949,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982634393041571e-05,
+      "loss": 1.7701,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982558228098771e-05,
+      "loss": 1.8765,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982482063155971e-05,
+      "loss": 1.7672,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98240589821317e-05,
+      "loss": 1.9316,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9823297332703706e-05,
+      "loss": 1.7627,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9822535683275704e-05,
+      "loss": 1.678,
+      "step": 2830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98217740338477e-05,
+      "loss": 1.9247,
+      "step": 2840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9821012384419705e-05,
+      "loss": 2.0473,
+      "step": 2850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98202507349917e-05,
+      "loss": 1.805,
+      "step": 2860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98194890855637e-05,
+      "loss": 1.8821,
+      "step": 2870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98187274361357e-05,
+      "loss": 1.7693,
+      "step": 2880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98179657867077e-05,
+      "loss": 1.6213,
+      "step": 2890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981720413727969e-05,
+      "loss": 1.8706,
+      "step": 2900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98164424878517e-05,
+      "loss": 1.7289,
+      "step": 2910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9815680838423694e-05,
+      "loss": 1.7842,
+      "step": 2920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981491918899569e-05,
+      "loss": 1.7525,
+      "step": 2930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981415753956769e-05,
+      "loss": 1.7481,
+      "step": 2940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9813395890139693e-05,
+      "loss": 1.8412,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9812634240711684e-05,
+      "loss": 1.6021,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.981187259128369e-05,
+      "loss": 2.0478,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9811110941855686e-05,
+      "loss": 1.9017,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.981034929242768e-05,
+      "loss": 1.9767,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980958764299968e-05,
+      "loss": 1.7376,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9808825993571685e-05,
+      "loss": 1.8679,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9808064344143676e-05,
+      "loss": 1.9908,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980730269471568e-05,
+      "loss": 1.7176,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980654104528768e-05,
+      "loss": 1.7423,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9805779395859675e-05,
+      "loss": 1.7206,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980501774643168e-05,
+      "loss": 1.6491,
+      "step": 3060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980425609700367e-05,
+      "loss": 1.8722,
+      "step": 3070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9803494447575674e-05,
+      "loss": 1.7412,
+      "step": 3080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980273279814767e-05,
+      "loss": 1.7331,
+      "step": 3090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980197114871967e-05,
+      "loss": 1.8489,
+      "step": 3100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9801209499291666e-05,
+      "loss": 1.6692,
+      "step": 3110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980044784986367e-05,
+      "loss": 1.7723,
+      "step": 3120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979968620043566e-05,
+      "loss": 1.8019,
+      "step": 3130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9798924551007666e-05,
+      "loss": 2.0198,
+      "step": 3140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979816290157966e-05,
+      "loss": 1.8323,
+      "step": 3150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979740125215166e-05,
+      "loss": 1.8626,
+      "step": 3160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979663960272366e-05,
+      "loss": 1.7609,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.979587795329566e-05,
+      "loss": 1.7569,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.979511630386765e-05,
+      "loss": 1.671,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.979435465443966e-05,
+      "loss": 1.6977,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9793593005011655e-05,
+      "loss": 1.8684,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.979283135558365e-05,
+      "loss": 1.704,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.979206970615565e-05,
+      "loss": 1.7804,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9791308056727654e-05,
+      "loss": 1.7963,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.979054640729965e-05,
+      "loss": 1.7375,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978978475787165e-05,
+      "loss": 1.8267,
+      "step": 3260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978902310844365e-05,
+      "loss": 1.6584,
+      "step": 3270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9788261459015644e-05,
+      "loss": 1.7467,
+      "step": 3280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978749980958765e-05,
+      "loss": 1.7374,
+      "step": 3290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9786738160159645e-05,
+      "loss": 1.7289,
+      "step": 3300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978597651073164e-05,
+      "loss": 1.7785,
+      "step": 3310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978521486130364e-05,
+      "loss": 2.0037,
+      "step": 3320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9784453211875645e-05,
+      "loss": 1.7621,
+      "step": 3330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9783691562447635e-05,
+      "loss": 1.6385,
+      "step": 3340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978292991301964e-05,
+      "loss": 1.755,
+      "step": 3350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978216826359164e-05,
+      "loss": 1.6991,
+      "step": 3360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9781406614163634e-05,
+      "loss": 1.8045,
+      "step": 3370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978064496473563e-05,
+      "loss": 1.6746,
+      "step": 3380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9779883315307636e-05,
+      "loss": 1.6609,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977912166587963e-05,
+      "loss": 1.7547,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977836001645163e-05,
+      "loss": 1.6422,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977759836702363e-05,
+      "loss": 1.7064,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9776836717595626e-05,
+      "loss": 1.7688,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9776075068167624e-05,
+      "loss": 1.7878,
+      "step": 3440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977531341873963e-05,
+      "loss": 2.1393,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9774551769311625e-05,
+      "loss": 1.6898,
+      "step": 3460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977379011988362e-05,
+      "loss": 1.5894,
+      "step": 3470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977302847045563e-05,
+      "loss": 1.6871,
+      "step": 3480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977226682102762e-05,
+      "loss": 1.7599,
+      "step": 3490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977150517159962e-05,
+      "loss": 1.7344,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977074352217161e-05,
+      "loss": 1.8123,
+      "step": 3510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.976998187274362e-05,
+      "loss": 1.4862,
+      "step": 3520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9769220223315614e-05,
+      "loss": 1.7748,
+      "step": 3530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.976845857388761e-05,
+      "loss": 1.7127,
+      "step": 3540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.976769692445961e-05,
+      "loss": 1.725,
+      "step": 3550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9766935275031613e-05,
+      "loss": 1.6842,
+      "step": 3560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9766173625603604e-05,
+      "loss": 1.9244,
+      "step": 3570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.976541197617561e-05,
+      "loss": 1.7457,
+      "step": 3580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9764650326747606e-05,
+      "loss": 1.6954,
+      "step": 3590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.97638886773196e-05,
+      "loss": 1.6902,
+      "step": 3600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.97631270278916e-05,
+      "loss": 1.65,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9762365378463605e-05,
+      "loss": 1.7247,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.97616037290356e-05,
+      "loss": 1.7447,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.97608420796076e-05,
+      "loss": 1.6846,
+      "step": 3640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9760080430179604e-05,
+      "loss": 1.7031,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9759318780751595e-05,
+      "loss": 1.6682,
+      "step": 3660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.97585571313236e-05,
+      "loss": 1.8842,
+      "step": 3670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.97577954818956e-05,
+      "loss": 1.8483,
+      "step": 3680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9757033832467594e-05,
+      "loss": 1.6491,
+      "step": 3690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.975627218303959e-05,
+      "loss": 1.684,
+      "step": 3700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9755510533611596e-05,
+      "loss": 1.5164,
+      "step": 3710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9754748884183586e-05,
+      "loss": 1.7094,
+      "step": 3720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.975398723475559e-05,
+      "loss": 1.655,
+      "step": 3730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.975322558532759e-05,
+      "loss": 1.6,
+      "step": 3740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9752463935899586e-05,
+      "loss": 1.5776,
+      "step": 3750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.975170228647158e-05,
+      "loss": 1.7506,
+      "step": 3760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.975094063704359e-05,
+      "loss": 1.5766,
+      "step": 3770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.975017898761558e-05,
+      "loss": 1.8097,
+      "step": 3780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.974941733818758e-05,
+      "loss": 1.6515,
+      "step": 3790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.974865568875958e-05,
+      "loss": 1.5615,
+      "step": 3800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.974789403933158e-05,
+      "loss": 1.6782,
+      "step": 3810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9747132389903575e-05,
+      "loss": 1.5652,
+      "step": 3820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.974637074047558e-05,
+      "loss": 1.686,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9745609091047576e-05,
+      "loss": 1.6738,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9744847441619574e-05,
+      "loss": 1.8036,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.974408579219158e-05,
+      "loss": 1.5509,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.974332414276357e-05,
+      "loss": 1.7174,
+      "step": 3870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.974256249333557e-05,
+      "loss": 1.7225,
+      "step": 3880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.974180084390757e-05,
+      "loss": 1.7589,
+      "step": 3890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.974103919447957e-05,
+      "loss": 1.7367,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9740277545051565e-05,
+      "loss": 1.657,
+      "step": 3910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973951589562357e-05,
+      "loss": 1.7899,
+      "step": 3920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973875424619556e-05,
+      "loss": 1.7264,
+      "step": 3930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9737992596767565e-05,
+      "loss": 1.6647,
+      "step": 3940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973723094733956e-05,
+      "loss": 1.6129,
+      "step": 3950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973646929791156e-05,
+      "loss": 1.6804,
+      "step": 3960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973570764848356e-05,
+      "loss": 1.4901,
+      "step": 3970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9734945999055555e-05,
+      "loss": 1.4965,
+      "step": 3980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973418434962755e-05,
+      "loss": 1.698,
+      "step": 3990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9733422700199556e-05,
+      "loss": 1.716,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9732661050771554e-05,
+      "loss": 1.5146,
+      "step": 4010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973189940134355e-05,
+      "loss": 1.7659,
+      "step": 4020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.973113775191555e-05,
+      "loss": 1.7332,
+      "step": 4030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9730376102487546e-05,
+      "loss": 1.7439,
+      "step": 4040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.972961445305955e-05,
+      "loss": 1.6838,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972885280363155e-05,
+      "loss": 1.73,
+      "step": 4060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9728091154203545e-05,
+      "loss": 1.728,
+      "step": 4070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972732950477554e-05,
+      "loss": 1.6675,
+      "step": 4080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972656785534755e-05,
+      "loss": 1.6273,
+      "step": 4090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972580620591954e-05,
+      "loss": 1.6244,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972504455649154e-05,
+      "loss": 1.7166,
+      "step": 4110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972428290706354e-05,
+      "loss": 1.6724,
+      "step": 4120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972352125763554e-05,
+      "loss": 1.6241,
+      "step": 4130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9722759608207534e-05,
+      "loss": 1.6969,
+      "step": 4140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972199795877954e-05,
+      "loss": 1.5854,
+      "step": 4150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.972123630935153e-05,
+      "loss": 1.7534,
+      "step": 4160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9720474659923533e-05,
+      "loss": 1.7051,
+      "step": 4170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971971301049553e-05,
+      "loss": 1.6349,
+      "step": 4180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971895136106753e-05,
+      "loss": 1.678,
+      "step": 4190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9718189711639526e-05,
+      "loss": 1.8842,
+      "step": 4200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971742806221153e-05,
+      "loss": 1.6166,
+      "step": 4210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971666641278353e-05,
+      "loss": 1.6829,
+      "step": 4220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9715904763355525e-05,
+      "loss": 1.5796,
+      "step": 4230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971514311392752e-05,
+      "loss": 1.6902,
+      "step": 4240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971438146449952e-05,
+      "loss": 1.5016,
+      "step": 4250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9713619815071524e-05,
+      "loss": 1.6283,
+      "step": 4260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.971285816564352e-05,
+      "loss": 1.6479,
+      "step": 4270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.971209651621552e-05,
+      "loss": 1.5963,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.971133486678752e-05,
+      "loss": 1.5781,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.971057321735952e-05,
+      "loss": 1.5803,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.970981156793151e-05,
+      "loss": 1.6222,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9709049918503516e-05,
+      "loss": 1.5437,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.970828826907551e-05,
+      "loss": 1.597,
+      "step": 4330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.970752661964751e-05,
+      "loss": 1.6402,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.970676497021951e-05,
+      "loss": 1.5925,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.970600332079151e-05,
+      "loss": 1.6829,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.97052416713635e-05,
+      "loss": 1.6079,
+      "step": 4370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.970448002193551e-05,
+      "loss": 1.5488,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9703718372507505e-05,
+      "loss": 1.6185,
+      "step": 4390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.97029567230795e-05,
+      "loss": 1.7197,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.97021950736515e-05,
+      "loss": 1.6796,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.97014334242235e-05,
+      "loss": 1.6951,
+      "step": 4420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.97006717747955e-05,
+      "loss": 1.6177,
+      "step": 4430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.96999101253675e-05,
+      "loss": 1.664,
+      "step": 4440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9699148475939496e-05,
+      "loss": 1.5584,
+      "step": 4450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9698386826511494e-05,
+      "loss": 1.6565,
+      "step": 4460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.96976251770835e-05,
+      "loss": 1.6395,
+      "step": 4470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.969686352765549e-05,
+      "loss": 1.6991,
+      "step": 4480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969610187822749e-05,
+      "loss": 1.5224,
+      "step": 4490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969534022879949e-05,
+      "loss": 1.5902,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969457857937149e-05,
+      "loss": 1.5924,
+      "step": 4510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9693816929943485e-05,
+      "loss": 1.7406,
+      "step": 4520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969305528051549e-05,
+      "loss": 1.5572,
+      "step": 4530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969229363108748e-05,
+      "loss": 1.6958,
+      "step": 4540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9691531981659485e-05,
+      "loss": 1.6348,
+      "step": 4550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969077033223148e-05,
+      "loss": 1.6531,
+      "step": 4560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.969000868280348e-05,
+      "loss": 1.7366,
+      "step": 4570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968924703337548e-05,
+      "loss": 1.5562,
+      "step": 4580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968848538394748e-05,
+      "loss": 1.6878,
+      "step": 4590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968772373451948e-05,
+      "loss": 1.6664,
+      "step": 4600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9686962085091476e-05,
+      "loss": 1.5303,
+      "step": 4610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9686200435663474e-05,
+      "loss": 1.6063,
+      "step": 4620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968543878623547e-05,
+      "loss": 1.6341,
+      "step": 4630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9684677136807475e-05,
+      "loss": 1.6236,
+      "step": 4640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968391548737947e-05,
+      "loss": 1.6974,
+      "step": 4650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968315383795147e-05,
+      "loss": 1.842,
+      "step": 4660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968239218852347e-05,
+      "loss": 1.6324,
+      "step": 4670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968163053909547e-05,
+      "loss": 1.7172,
+      "step": 4680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968086888966746e-05,
+      "loss": 1.5984,
+      "step": 4690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.968010724023947e-05,
+      "loss": 1.6541,
+      "step": 4700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9679345590811464e-05,
+      "loss": 1.6534,
+      "step": 4710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967858394138346e-05,
+      "loss": 1.7619,
+      "step": 4720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967782229195546e-05,
+      "loss": 1.7375,
+      "step": 4730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9677060642527464e-05,
+      "loss": 1.7461,
+      "step": 4740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9676298993099454e-05,
+      "loss": 1.6525,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967553734367146e-05,
+      "loss": 1.5027,
+      "step": 4760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9674775694243456e-05,
+      "loss": 1.6237,
+      "step": 4770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9674014044815453e-05,
+      "loss": 1.5023,
+      "step": 4780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967325239538745e-05,
+      "loss": 1.383,
+      "step": 4790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9672490745959455e-05,
+      "loss": 1.7156,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967172909653145e-05,
+      "loss": 1.6065,
+      "step": 4810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967096744710345e-05,
+      "loss": 1.4807,
+      "step": 4820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.967020579767545e-05,
+      "loss": 1.6201,
+      "step": 4830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9669444148247445e-05,
+      "loss": 1.5722,
+      "step": 4840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966868249881945e-05,
+      "loss": 1.5964,
+      "step": 4850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966792084939145e-05,
+      "loss": 1.6957,
+      "step": 4860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9667159199963444e-05,
+      "loss": 1.587,
+      "step": 4870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966639755053544e-05,
+      "loss": 1.5629,
+      "step": 4880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966563590110744e-05,
+      "loss": 1.5512,
+      "step": 4890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966487425167944e-05,
+      "loss": 1.6292,
+      "step": 4900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966411260225144e-05,
+      "loss": 1.6129,
+      "step": 4910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.966335095282343e-05,
+      "loss": 1.607,
+      "step": 4920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9662589303395436e-05,
+      "loss": 1.6512,
+      "step": 4930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.966182765396743e-05,
+      "loss": 1.7334,
+      "step": 4940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.966106600453943e-05,
+      "loss": 1.432,
+      "step": 4950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.966030435511143e-05,
+      "loss": 1.7051,
+      "step": 4960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965954270568343e-05,
+      "loss": 1.4735,
+      "step": 4970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965878105625543e-05,
+      "loss": 1.659,
+      "step": 4980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965801940682743e-05,
+      "loss": 1.5303,
+      "step": 4990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9657257757399425e-05,
+      "loss": 1.6354,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965649610797142e-05,
+      "loss": 1.807,
+      "step": 5010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9655734458543427e-05,
+      "loss": 1.7377,
+      "step": 5020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9654972809115424e-05,
+      "loss": 1.4959,
+      "step": 5030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965421115968742e-05,
+      "loss": 1.6932,
+      "step": 5040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965344951025942e-05,
+      "loss": 1.4589,
+      "step": 5050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965268786083142e-05,
+      "loss": 1.5529,
+      "step": 5060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9651926211403414e-05,
+      "loss": 1.5537,
+      "step": 5070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.965116456197542e-05,
+      "loss": 1.678,
+      "step": 5080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9650402912547416e-05,
+      "loss": 1.5997,
+      "step": 5090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.964964126311941e-05,
+      "loss": 1.5754,
+      "step": 5100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.964887961369141e-05,
+      "loss": 1.4722,
+      "step": 5110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9648117964263415e-05,
+      "loss": 1.8217,
+      "step": 5120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9647356314835405e-05,
+      "loss": 1.8546,
+      "step": 5130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.964659466540741e-05,
+      "loss": 1.6714,
+      "step": 5140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.964583301597941e-05,
+      "loss": 1.701,
+      "step": 5150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9645071366551405e-05,
+      "loss": 1.555,
+      "step": 5160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96443097171234e-05,
+      "loss": 1.5997,
+      "step": 5170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9643548067695406e-05,
+      "loss": 1.7099,
+      "step": 5180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9642786418267404e-05,
+      "loss": 1.599,
+      "step": 5190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96420247688394e-05,
+      "loss": 1.647,
+      "step": 5200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96412631194114e-05,
+      "loss": 1.5166,
+      "step": 5210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9640501469983396e-05,
+      "loss": 1.5513,
+      "step": 5220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96397398205554e-05,
+      "loss": 1.4816,
+      "step": 5230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96389781711274e-05,
+      "loss": 1.5121,
+      "step": 5240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9638216521699395e-05,
+      "loss": 1.6454,
+      "step": 5250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963745487227139e-05,
+      "loss": 1.5571,
+      "step": 5260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96366932228434e-05,
+      "loss": 1.492,
+      "step": 5270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963593157341539e-05,
+      "loss": 1.5923,
+      "step": 5280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963516992398739e-05,
+      "loss": 1.6293,
+      "step": 5290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963440827455939e-05,
+      "loss": 1.4573,
+      "step": 5300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963364662513139e-05,
+      "loss": 1.5389,
+      "step": 5310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9632884975703384e-05,
+      "loss": 1.5586,
+      "step": 5320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963212332627539e-05,
+      "loss": 1.5391,
+      "step": 5330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.963136167684738e-05,
+      "loss": 1.4421,
+      "step": 5340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9630600027419384e-05,
+      "loss": 1.7778,
+      "step": 5350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.962983837799138e-05,
+      "loss": 1.5606,
+      "step": 5360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.962907672856338e-05,
+      "loss": 1.5499,
+      "step": 5370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9628315079135376e-05,
+      "loss": 1.5715,
+      "step": 5380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9627553429707374e-05,
+      "loss": 1.6444,
+      "step": 5390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.962679178027938e-05,
+      "loss": 1.472,
+      "step": 5400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9626030130851375e-05,
+      "loss": 1.6052,
+      "step": 5410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.962526848142337e-05,
+      "loss": 1.5152,
+      "step": 5420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.962450683199537e-05,
+      "loss": 1.4433,
+      "step": 5430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9623745182567374e-05,
+      "loss": 1.7108,
+      "step": 5440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9622983533139365e-05,
+      "loss": 1.5802,
+      "step": 5450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.962222188371137e-05,
+      "loss": 1.5738,
+      "step": 5460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.962146023428337e-05,
+      "loss": 1.5699,
+      "step": 5470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9620698584855364e-05,
+      "loss": 1.6516,
+      "step": 5480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961993693542736e-05,
+      "loss": 1.5635,
+      "step": 5490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9619175285999366e-05,
+      "loss": 1.5381,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961841363657136e-05,
+      "loss": 1.5924,
+      "step": 5510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961765198714336e-05,
+      "loss": 1.6525,
+      "step": 5520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961689033771536e-05,
+      "loss": 1.6992,
+      "step": 5530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9616128688287356e-05,
+      "loss": 1.6165,
+      "step": 5540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961536703885935e-05,
+      "loss": 1.5774,
+      "step": 5550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961460538943136e-05,
+      "loss": 1.6195,
+      "step": 5560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9613843740003355e-05,
+      "loss": 1.6283,
+      "step": 5570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961308209057535e-05,
+      "loss": 1.6055,
+      "step": 5580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.961232044114735e-05,
+      "loss": 1.5324,
+      "step": 5590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.961155879171935e-05,
+      "loss": 1.6448,
+      "step": 5600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.961079714229135e-05,
+      "loss": 1.5065,
+      "step": 5610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.961003549286335e-05,
+      "loss": 1.3574,
+      "step": 5620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9609273843435347e-05,
+      "loss": 1.4543,
+      "step": 5630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9608512194007344e-05,
+      "loss": 1.6374,
+      "step": 5640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960775054457935e-05,
+      "loss": 1.7077,
+      "step": 5650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960698889515134e-05,
+      "loss": 1.4284,
+      "step": 5660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960622724572334e-05,
+      "loss": 1.5947,
+      "step": 5670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960546559629534e-05,
+      "loss": 1.686,
+      "step": 5680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960470394686734e-05,
+      "loss": 1.558,
+      "step": 5690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9603942297439336e-05,
+      "loss": 1.552,
+      "step": 5700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960318064801134e-05,
+      "loss": 1.4332,
+      "step": 5710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960241899858333e-05,
+      "loss": 1.4902,
+      "step": 5720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9601657349155335e-05,
+      "loss": 1.4964,
+      "step": 5730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960089569972733e-05,
+      "loss": 1.628,
+      "step": 5740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.960013405029933e-05,
+      "loss": 1.7822,
+      "step": 5750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.959937240087133e-05,
+      "loss": 1.4517,
+      "step": 5760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.959861075144333e-05,
+      "loss": 1.5984,
+      "step": 5770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.959784910201533e-05,
+      "loss": 1.6394,
+      "step": 5780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9597087452587326e-05,
+      "loss": 1.6932,
+      "step": 5790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9596325803159324e-05,
+      "loss": 1.3916,
+      "step": 5800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.959556415373132e-05,
+      "loss": 1.4691,
+      "step": 5810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9594802504303326e-05,
+      "loss": 1.4779,
+      "step": 5820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9594040854875316e-05,
+      "loss": 1.6395,
+      "step": 5830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.959327920544732e-05,
+      "loss": 1.5017,
+      "step": 5840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.959251755601932e-05,
+      "loss": 1.5928,
+      "step": 5850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9591755906591315e-05,
+      "loss": 1.6306,
+      "step": 5860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.959099425716331e-05,
+      "loss": 1.5162,
+      "step": 5870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.959023260773532e-05,
+      "loss": 1.7551,
+      "step": 5880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.958947095830731e-05,
+      "loss": 1.6325,
+      "step": 5890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.958870930887931e-05,
+      "loss": 1.529,
+      "step": 5900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.958794765945131e-05,
+      "loss": 1.4474,
+      "step": 5910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.958718601002331e-05,
+      "loss": 1.5045,
+      "step": 5920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9586424360595304e-05,
+      "loss": 1.5197,
+      "step": 5930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.958566271116731e-05,
+      "loss": 1.5708,
+      "step": 5940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9584901061739306e-05,
+      "loss": 1.5372,
+      "step": 5950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9584139412311304e-05,
+      "loss": 1.6256,
+      "step": 5960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95833777628833e-05,
+      "loss": 1.4375,
+      "step": 5970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95826161134553e-05,
+      "loss": 1.5866,
+      "step": 5980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95818544640273e-05,
+      "loss": 1.5366,
+      "step": 5990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95810928145993e-05,
+      "loss": 1.505,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95803311651713e-05,
+      "loss": 1.4372,
+      "step": 6010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9579569515743295e-05,
+      "loss": 1.5738,
+      "step": 6020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.95788078663153e-05,
+      "loss": 1.5582,
+      "step": 6030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957804621688729e-05,
+      "loss": 1.4368,
+      "step": 6040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9577284567459294e-05,
+      "loss": 1.6094,
+      "step": 6050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957652291803129e-05,
+      "loss": 1.491,
+      "step": 6060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957576126860329e-05,
+      "loss": 1.4423,
+      "step": 6070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957499961917529e-05,
+      "loss": 1.6031,
+      "step": 6080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957423796974729e-05,
+      "loss": 1.6001,
+      "step": 6090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957347632031928e-05,
+      "loss": 1.6379,
+      "step": 6100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9572714670891286e-05,
+      "loss": 1.4268,
+      "step": 6110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9571953021463283e-05,
+      "loss": 1.4268,
+      "step": 6120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957119137203528e-05,
+      "loss": 1.425,
+      "step": 6130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.957042972260728e-05,
+      "loss": 1.7196,
+      "step": 6140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956966807317928e-05,
+      "loss": 1.6246,
+      "step": 6150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956890642375128e-05,
+      "loss": 1.5407,
+      "step": 6160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956814477432328e-05,
+      "loss": 1.4594,
+      "step": 6170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9567383124895275e-05,
+      "loss": 1.5897,
+      "step": 6180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956662147546727e-05,
+      "loss": 1.4055,
+      "step": 6190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956585982603928e-05,
+      "loss": 1.5106,
+      "step": 6200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9565098176611274e-05,
+      "loss": 1.4789,
+      "step": 6210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956433652718327e-05,
+      "loss": 1.621,
+      "step": 6220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956357487775527e-05,
+      "loss": 1.602,
+      "step": 6230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.956281322832727e-05,
+      "loss": 1.642,
+      "step": 6240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9562051578899264e-05,
+      "loss": 1.5103,
+      "step": 6250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.956128992947127e-05,
+      "loss": 1.2975,
+      "step": 6260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.956052828004326e-05,
+      "loss": 1.4497,
+      "step": 6270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955976663061526e-05,
+      "loss": 1.4488,
+      "step": 6280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955900498118726e-05,
+      "loss": 1.6152,
+      "step": 6290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955824333175926e-05,
+      "loss": 1.4715,
+      "step": 6300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9557481682331256e-05,
+      "loss": 1.5672,
+      "step": 6310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955672003290326e-05,
+      "loss": 1.5298,
+      "step": 6320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955595838347525e-05,
+      "loss": 1.411,
+      "step": 6330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9555196734047255e-05,
+      "loss": 1.6484,
+      "step": 6340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955443508461925e-05,
+      "loss": 1.4544,
+      "step": 6350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955367343519125e-05,
+      "loss": 1.3545,
+      "step": 6360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9552911785763254e-05,
+      "loss": 1.5908,
+      "step": 6370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955215013633525e-05,
+      "loss": 1.5166,
+      "step": 6380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.955138848690725e-05,
+      "loss": 1.5503,
+      "step": 6390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9550626837479246e-05,
+      "loss": 1.4337,
+      "step": 6400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.954986518805125e-05,
+      "loss": 1.4866,
+      "step": 6410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.954910353862324e-05,
+      "loss": 1.6988,
+      "step": 6420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9548341889195246e-05,
+      "loss": 1.5241,
+      "step": 6430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.954758023976724e-05,
+      "loss": 1.4776,
+      "step": 6440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.954681859033924e-05,
+      "loss": 1.4041,
+      "step": 6450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.954605694091124e-05,
+      "loss": 1.4536,
+      "step": 6460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.954529529148324e-05,
+      "loss": 1.4259,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.954453364205523e-05,
+      "loss": 1.5767,
+      "step": 6480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.954377199262724e-05,
+      "loss": 1.4998,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9543010343199235e-05,
+      "loss": 1.3588,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.954224869377123e-05,
+      "loss": 1.7313,
+      "step": 6510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.954148704434323e-05,
+      "loss": 1.5771,
+      "step": 6520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9540725394915234e-05,
+      "loss": 1.491,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9539963745487224e-05,
+      "loss": 1.3665,
+      "step": 6540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953920209605923e-05,
+      "loss": 1.4204,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9538440446631226e-05,
+      "loss": 1.5955,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9537678797203224e-05,
+      "loss": 1.6196,
+      "step": 6570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953691714777523e-05,
+      "loss": 1.6907,
+      "step": 6580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9536155498347225e-05,
+      "loss": 1.4794,
+      "step": 6590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953539384891922e-05,
+      "loss": 1.5612,
+      "step": 6600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953463219949122e-05,
+      "loss": 1.51,
+      "step": 6610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9533870550063225e-05,
+      "loss": 1.386,
+      "step": 6620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9533108900635215e-05,
+      "loss": 1.4816,
+      "step": 6630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953234725120722e-05,
+      "loss": 1.4524,
+      "step": 6640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953158560177922e-05,
+      "loss": 1.4866,
+      "step": 6650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9530823952351214e-05,
+      "loss": 1.4965,
+      "step": 6660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.953006230292321e-05,
+      "loss": 1.4886,
+      "step": 6670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9529300653495216e-05,
+      "loss": 1.6532,
+      "step": 6680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.952853900406721e-05,
+      "loss": 1.5153,
+      "step": 6690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.952777735463921e-05,
+      "loss": 1.3405,
+      "step": 6700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.952701570521121e-05,
+      "loss": 1.4654,
+      "step": 6710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9526254055783206e-05,
+      "loss": 1.3844,
+      "step": 6720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9525492406355203e-05,
+      "loss": 1.5712,
+      "step": 6730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95247307569272e-05,
+      "loss": 1.4924,
+      "step": 6740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9523969107499205e-05,
+      "loss": 1.3982,
+      "step": 6750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95232074580712e-05,
+      "loss": 1.44,
+      "step": 6760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95224458086432e-05,
+      "loss": 1.5073,
+      "step": 6770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95216841592152e-05,
+      "loss": 1.4486,
+      "step": 6780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95209225097872e-05,
+      "loss": 1.512,
+      "step": 6790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.952016086035919e-05,
+      "loss": 1.5749,
+      "step": 6800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95193992109312e-05,
+      "loss": 1.602,
+      "step": 6810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9518637561503194e-05,
+      "loss": 1.375,
+      "step": 6820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.951787591207519e-05,
+      "loss": 1.4885,
+      "step": 6830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.951711426264719e-05,
+      "loss": 1.4365,
+      "step": 6840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.951635261321919e-05,
+      "loss": 1.4765,
+      "step": 6850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9515590963791184e-05,
+      "loss": 1.6909,
+      "step": 6860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.951482931436319e-05,
+      "loss": 1.4077,
+      "step": 6870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9514067664935186e-05,
+      "loss": 1.3866,
+      "step": 6880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.951330601550718e-05,
+      "loss": 1.5315,
+      "step": 6890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.951254436607918e-05,
+      "loss": 1.6106,
+      "step": 6900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9511782716651185e-05,
+      "loss": 1.4799,
+      "step": 6910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9511021067223176e-05,
+      "loss": 1.6922,
+      "step": 6920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.951025941779518e-05,
+      "loss": 1.497,
+      "step": 6930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950949776836718e-05,
+      "loss": 1.6078,
+      "step": 6940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9508736118939175e-05,
+      "loss": 1.5296,
+      "step": 6950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950797446951118e-05,
+      "loss": 1.4922,
+      "step": 6960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9507212820083177e-05,
+      "loss": 1.4675,
+      "step": 6970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9506451170655174e-05,
+      "loss": 1.524,
+      "step": 6980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950568952122717e-05,
+      "loss": 1.4638,
+      "step": 6990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9504927871799176e-05,
+      "loss": 1.4654,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9504166222371166e-05,
+      "loss": 1.4517,
+      "step": 7010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950340457294317e-05,
+      "loss": 1.367,
+      "step": 7020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950264292351517e-05,
+      "loss": 1.527,
+      "step": 7030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9501881274087166e-05,
+      "loss": 1.4645,
+      "step": 7040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950111962465916e-05,
+      "loss": 1.4524,
+      "step": 7050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950035797523117e-05,
+      "loss": 1.3867,
+      "step": 7060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949959632580316e-05,
+      "loss": 1.3662,
+      "step": 7070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949883467637516e-05,
+      "loss": 1.4959,
+      "step": 7080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949807302694716e-05,
+      "loss": 1.4981,
+      "step": 7090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949731137751916e-05,
+      "loss": 1.5869,
+      "step": 7100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9496549728091155e-05,
+      "loss": 1.6878,
+      "step": 7110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949578807866316e-05,
+      "loss": 1.6248,
+      "step": 7120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949502642923515e-05,
+      "loss": 1.4251,
+      "step": 7130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9494264779807154e-05,
+      "loss": 1.713,
+      "step": 7140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949350313037915e-05,
+      "loss": 1.6612,
+      "step": 7150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949274148095115e-05,
+      "loss": 1.4881,
+      "step": 7160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949197983152315e-05,
+      "loss": 1.5939,
+      "step": 7170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949121818209515e-05,
+      "loss": 1.4313,
+      "step": 7180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.949045653266715e-05,
+      "loss": 1.3092,
+      "step": 7190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9489694883239145e-05,
+      "loss": 1.4968,
+      "step": 7200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948893323381114e-05,
+      "loss": 1.3771,
+      "step": 7210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948817158438314e-05,
+      "loss": 1.5074,
+      "step": 7220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9487409934955145e-05,
+      "loss": 1.4254,
+      "step": 7230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9486648285527135e-05,
+      "loss": 1.3878,
+      "step": 7240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948588663609914e-05,
+      "loss": 1.5459,
+      "step": 7250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948512498667114e-05,
+      "loss": 1.5382,
+      "step": 7260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9484363337243134e-05,
+      "loss": 1.5585,
+      "step": 7270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948360168781513e-05,
+      "loss": 1.4538,
+      "step": 7280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9482840038387136e-05,
+      "loss": 1.5015,
+      "step": 7290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948207838895913e-05,
+      "loss": 1.7403,
+      "step": 7300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948131673953113e-05,
+      "loss": 1.5183,
+      "step": 7310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948055509010313e-05,
+      "loss": 1.6055,
+      "step": 7320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9479793440675126e-05,
+      "loss": 1.4813,
+      "step": 7330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9479031791247123e-05,
+      "loss": 1.3896,
+      "step": 7340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947827014181913e-05,
+      "loss": 1.4216,
+      "step": 7350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9477508492391125e-05,
+      "loss": 1.6834,
+      "step": 7360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947674684296312e-05,
+      "loss": 1.3757,
+      "step": 7370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947598519353513e-05,
+      "loss": 1.5731,
+      "step": 7380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947522354410712e-05,
+      "loss": 1.6369,
+      "step": 7390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947446189467912e-05,
+      "loss": 1.2534,
+      "step": 7400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947370024525112e-05,
+      "loss": 1.3786,
+      "step": 7410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947293859582312e-05,
+      "loss": 1.6177,
+      "step": 7420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9472176946395114e-05,
+      "loss": 1.5847,
+      "step": 7430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947141529696712e-05,
+      "loss": 1.4073,
+      "step": 7440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.947065364753911e-05,
+      "loss": 1.3373,
+      "step": 7450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9469891998111113e-05,
+      "loss": 1.6774,
+      "step": 7460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.946913034868311e-05,
+      "loss": 1.3398,
+      "step": 7470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.946836869925511e-05,
+      "loss": 1.5537,
+      "step": 7480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9467607049827106e-05,
+      "loss": 1.3872,
+      "step": 7490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.946684540039911e-05,
+      "loss": 1.5859,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.94660837509711e-05,
+      "loss": 1.4365,
+      "step": 7510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9465322101543105e-05,
+      "loss": 1.361,
+      "step": 7520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.94645604521151e-05,
+      "loss": 1.6832,
+      "step": 7530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.94637988026871e-05,
+      "loss": 1.4872,
+      "step": 7540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9463037153259104e-05,
+      "loss": 1.3692,
+      "step": 7550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.94622755038311e-05,
+      "loss": 1.4837,
+      "step": 7560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.94615138544031e-05,
+      "loss": 1.4386,
+      "step": 7570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9460752204975097e-05,
+      "loss": 1.4892,
+      "step": 7580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.94599905555471e-05,
+      "loss": 1.3251,
+      "step": 7590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945922890611909e-05,
+      "loss": 1.413,
+      "step": 7600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9458467256691096e-05,
+      "loss": 1.4397,
+      "step": 7610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945770560726309e-05,
+      "loss": 1.3452,
+      "step": 7620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945694395783509e-05,
+      "loss": 1.4522,
+      "step": 7630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945618230840709e-05,
+      "loss": 1.4811,
+      "step": 7640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9455420658979086e-05,
+      "loss": 1.5733,
+      "step": 7650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945465900955108e-05,
+      "loss": 1.5046,
+      "step": 7660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945389736012309e-05,
+      "loss": 1.4072,
+      "step": 7670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945313571069508e-05,
+      "loss": 1.6471,
+      "step": 7680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945237406126708e-05,
+      "loss": 1.4507,
+      "step": 7690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945161241183908e-05,
+      "loss": 1.4277,
+      "step": 7700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.945085076241108e-05,
+      "loss": 1.456,
+      "step": 7710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9450089112983075e-05,
+      "loss": 1.5445,
+      "step": 7720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.944932746355508e-05,
+      "loss": 1.4754,
+      "step": 7730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9448565814127076e-05,
+      "loss": 1.5327,
+      "step": 7740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9447804164699074e-05,
+      "loss": 1.4829,
+      "step": 7750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.944704251527108e-05,
+      "loss": 1.6795,
+      "step": 7760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.944628086584307e-05,
+      "loss": 1.4683,
+      "step": 7770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.944551921641507e-05,
+      "loss": 1.581,
+      "step": 7780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.944475756698707e-05,
+      "loss": 1.6855,
+      "step": 7790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.944399591755907e-05,
+      "loss": 1.6085,
+      "step": 7800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9443234268131065e-05,
+      "loss": 1.4563,
+      "step": 7810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.944247261870307e-05,
+      "loss": 1.3233,
+      "step": 7820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.944171096927506e-05,
+      "loss": 1.5191,
+      "step": 7830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9440949319847065e-05,
+      "loss": 1.4151,
+      "step": 7840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.944018767041906e-05,
+      "loss": 1.3591,
+      "step": 7850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943942602099106e-05,
+      "loss": 1.6094,
+      "step": 7860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943866437156306e-05,
+      "loss": 1.3702,
+      "step": 7870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943790272213506e-05,
+      "loss": 1.4132,
+      "step": 7880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943714107270705e-05,
+      "loss": 1.4552,
+      "step": 7890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9436379423279056e-05,
+      "loss": 1.5437,
+      "step": 7900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9435617773851054e-05,
+      "loss": 1.3571,
+      "step": 7910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943485612442305e-05,
+      "loss": 1.3626,
+      "step": 7920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943409447499505e-05,
+      "loss": 1.4289,
+      "step": 7930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943333282556705e-05,
+      "loss": 1.4231,
+      "step": 7940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943257117613905e-05,
+      "loss": 1.4423,
+      "step": 7950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943180952671105e-05,
+      "loss": 1.3791,
+      "step": 7960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943104787728305e-05,
+      "loss": 1.4959,
+      "step": 7970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.943028622785504e-05,
+      "loss": 1.4397,
+      "step": 7980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.942952457842705e-05,
+      "loss": 1.4879,
+      "step": 7990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9428762928999044e-05,
+      "loss": 1.5009,
+      "step": 8000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.942800127957104e-05,
+      "loss": 1.4805,
+      "step": 8010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.942723963014304e-05,
+      "loss": 1.4159,
+      "step": 8020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9426477980715044e-05,
+      "loss": 1.5213,
+      "step": 8030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9425716331287034e-05,
+      "loss": 1.5424,
+      "step": 8040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.942495468185904e-05,
+      "loss": 1.4818,
+      "step": 8050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9424193032431036e-05,
+      "loss": 1.511,
+      "step": 8060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9423431383003033e-05,
+      "loss": 1.3838,
+      "step": 8070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.942266973357503e-05,
+      "loss": 1.4245,
+      "step": 8080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9421908084147035e-05,
+      "loss": 1.5944,
+      "step": 8090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9421146434719026e-05,
+      "loss": 1.4024,
+      "step": 8100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.942038478529103e-05,
+      "loss": 1.4239,
+      "step": 8110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941962313586303e-05,
+      "loss": 1.542,
+      "step": 8120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9418861486435025e-05,
+      "loss": 1.5723,
+      "step": 8130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941809983700702e-05,
+      "loss": 1.4905,
+      "step": 8140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941733818757902e-05,
+      "loss": 1.53,
+      "step": 8150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9416576538151024e-05,
+      "loss": 1.2325,
+      "step": 8160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941581488872302e-05,
+      "loss": 1.3458,
+      "step": 8170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941505323929502e-05,
+      "loss": 1.5492,
+      "step": 8180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9414291589867017e-05,
+      "loss": 1.4129,
+      "step": 8190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941352994043902e-05,
+      "loss": 1.4233,
+      "step": 8200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.941276829101101e-05,
+      "loss": 1.2377,
+      "step": 8210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9412006641583016e-05,
+      "loss": 1.4066,
+      "step": 8220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.941124499215501e-05,
+      "loss": 1.4305,
+      "step": 8230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.941048334272701e-05,
+      "loss": 1.4324,
+      "step": 8240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.940972169329901e-05,
+      "loss": 1.4575,
+      "step": 8250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.940896004387101e-05,
+      "loss": 1.4481,
+      "step": 8260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9408198394443e-05,
+      "loss": 1.3935,
+      "step": 8270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.940743674501501e-05,
+      "loss": 1.3854,
+      "step": 8280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9406675095587005e-05,
+      "loss": 1.3787,
+      "step": 8290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9405913446159e-05,
+      "loss": 1.4449,
+      "step": 8300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9405151796731e-05,
+      "loss": 1.4391,
+      "step": 8310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9404390147303004e-05,
+      "loss": 1.4161,
+      "step": 8320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9403628497875e-05,
+      "loss": 1.5529,
+      "step": 8330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9402866848447e-05,
+      "loss": 1.4841,
+      "step": 8340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9402105199019e-05,
+      "loss": 1.3331,
+      "step": 8350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9401343549590994e-05,
+      "loss": 1.4807,
+      "step": 8360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9400581900163e-05,
+      "loss": 1.5402,
+      "step": 8370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9399820250734996e-05,
+      "loss": 1.5372,
+      "step": 8380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.939905860130699e-05,
+      "loss": 1.5184,
+      "step": 8390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.939829695187899e-05,
+      "loss": 1.5174,
+      "step": 8400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9397535302450995e-05,
+      "loss": 1.3865,
+      "step": 8410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9396773653022985e-05,
+      "loss": 1.3069,
+      "step": 8420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.939601200359499e-05,
+      "loss": 1.375,
+      "step": 8430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.939525035416699e-05,
+      "loss": 1.4485,
+      "step": 8440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9394488704738985e-05,
+      "loss": 1.3709,
+      "step": 8450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.939372705531098e-05,
+      "loss": 1.4909,
+      "step": 8460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9392965405882986e-05,
+      "loss": 1.4936,
+      "step": 8470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.939220375645498e-05,
+      "loss": 1.5602,
+      "step": 8480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.939144210702698e-05,
+      "loss": 1.427,
+      "step": 8490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.939068045759898e-05,
+      "loss": 1.3926,
+      "step": 8500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9389918808170976e-05,
+      "loss": 1.5437,
+      "step": 8510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9389157158742974e-05,
+      "loss": 1.4765,
+      "step": 8520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938839550931498e-05,
+      "loss": 1.5177,
+      "step": 8530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9387633859886975e-05,
+      "loss": 1.4327,
+      "step": 8540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938687221045897e-05,
+      "loss": 1.6322,
+      "step": 8550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938611056103098e-05,
+      "loss": 1.3167,
+      "step": 8560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938534891160297e-05,
+      "loss": 1.4539,
+      "step": 8570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938458726217497e-05,
+      "loss": 1.4382,
+      "step": 8580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938382561274696e-05,
+      "loss": 1.4558,
+      "step": 8590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938306396331897e-05,
+      "loss": 1.6293,
+      "step": 8600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9382302313890964e-05,
+      "loss": 1.554,
+      "step": 8610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938154066446296e-05,
+      "loss": 1.4785,
+      "step": 8620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.938077901503496e-05,
+      "loss": 1.4706,
+      "step": 8630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9380017365606964e-05,
+      "loss": 1.4369,
+      "step": 8640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9379255716178954e-05,
+      "loss": 1.507,
+      "step": 8650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937849406675096e-05,
+      "loss": 1.6385,
+      "step": 8660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9377732417322956e-05,
+      "loss": 1.3839,
+      "step": 8670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9376970767894953e-05,
+      "loss": 1.6495,
+      "step": 8680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937620911846695e-05,
+      "loss": 1.3931,
+      "step": 8690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9375447469038955e-05,
+      "loss": 1.4902,
+      "step": 8700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937468581961095e-05,
+      "loss": 1.4762,
+      "step": 8710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937392417018295e-05,
+      "loss": 1.4095,
+      "step": 8720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937316252075495e-05,
+      "loss": 1.4149,
+      "step": 8730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9372400871326945e-05,
+      "loss": 1.4997,
+      "step": 8740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937163922189895e-05,
+      "loss": 1.4546,
+      "step": 8750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.937087757247095e-05,
+      "loss": 1.3472,
+      "step": 8760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9370115923042944e-05,
+      "loss": 1.5443,
+      "step": 8770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.936935427361494e-05,
+      "loss": 1.458,
+      "step": 8780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9368592624186946e-05,
+      "loss": 1.375,
+      "step": 8790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9367830974758937e-05,
+      "loss": 1.542,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.936706932533094e-05,
+      "loss": 1.4439,
+      "step": 8810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.936630767590294e-05,
+      "loss": 1.4427,
+      "step": 8820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9365546026474936e-05,
+      "loss": 1.5113,
+      "step": 8830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.936478437704693e-05,
+      "loss": 1.3747,
+      "step": 8840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.936402272761894e-05,
+      "loss": 1.5697,
+      "step": 8850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.936326107819093e-05,
+      "loss": 1.4358,
+      "step": 8860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.936249942876293e-05,
+      "loss": 1.3581,
+      "step": 8870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.936173777933493e-05,
+      "loss": 1.5213,
+      "step": 8880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.936097612990693e-05,
+      "loss": 1.3443,
+      "step": 8890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9360214480478925e-05,
+      "loss": 1.3839,
+      "step": 8900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935945283105093e-05,
+      "loss": 1.5222,
+      "step": 8910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9358691181622926e-05,
+      "loss": 1.3619,
+      "step": 8920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9357929532194924e-05,
+      "loss": 1.4335,
+      "step": 8930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935716788276693e-05,
+      "loss": 1.4594,
+      "step": 8940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935640623333892e-05,
+      "loss": 1.7465,
+      "step": 8950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935564458391092e-05,
+      "loss": 1.3742,
+      "step": 8960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935488293448292e-05,
+      "loss": 1.5442,
+      "step": 8970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935412128505492e-05,
+      "loss": 1.4644,
+      "step": 8980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9353359635626916e-05,
+      "loss": 1.4326,
+      "step": 8990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935259798619892e-05,
+      "loss": 1.4701,
+      "step": 9000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.935183633677091e-05,
+      "loss": 1.5854,
+      "step": 9010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9351074687342915e-05,
+      "loss": 1.4573,
+      "step": 9020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9350313037914905e-05,
+      "loss": 1.3243,
+      "step": 9030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.934955138848691e-05,
+      "loss": 1.6106,
+      "step": 9040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.934878973905891e-05,
+      "loss": 1.3277,
+      "step": 9050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9348028089630905e-05,
+      "loss": 1.4677,
+      "step": 9060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.93472664402029e-05,
+      "loss": 1.4452,
+      "step": 9070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9346504790774906e-05,
+      "loss": 1.7204,
+      "step": 9080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9345743141346904e-05,
+      "loss": 1.6249,
+      "step": 9090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.93449814919189e-05,
+      "loss": 1.4378,
+      "step": 9100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.93442198424909e-05,
+      "loss": 1.6146,
+      "step": 9110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9343458193062896e-05,
+      "loss": 1.3419,
+      "step": 9120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.93426965436349e-05,
+      "loss": 1.3795,
+      "step": 9130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.93419348942069e-05,
+      "loss": 1.4137,
+      "step": 9140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9341173244778895e-05,
+      "loss": 1.2591,
+      "step": 9150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.934041159535089e-05,
+      "loss": 1.5363,
+      "step": 9160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.93396499459229e-05,
+      "loss": 1.3919,
+      "step": 9170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933888829649489e-05,
+      "loss": 1.4032,
+      "step": 9180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933812664706689e-05,
+      "loss": 1.4208,
+      "step": 9190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933736499763889e-05,
+      "loss": 1.4606,
+      "step": 9200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933660334821089e-05,
+      "loss": 1.5342,
+      "step": 9210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9335841698782884e-05,
+      "loss": 1.4319,
+      "step": 9220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933508004935489e-05,
+      "loss": 1.4659,
+      "step": 9230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933431839992688e-05,
+      "loss": 1.4699,
+      "step": 9240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9333556750498884e-05,
+      "loss": 1.3317,
+      "step": 9250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933279510107088e-05,
+      "loss": 1.303,
+      "step": 9260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933203345164288e-05,
+      "loss": 1.3177,
+      "step": 9270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9331271802214876e-05,
+      "loss": 1.4764,
+      "step": 9280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.933051015278688e-05,
+      "loss": 1.4683,
+      "step": 9290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.932974850335888e-05,
+      "loss": 1.3229,
+      "step": 9300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9328986853930875e-05,
+      "loss": 1.5648,
+      "step": 9310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932822520450287e-05,
+      "loss": 1.4743,
+      "step": 9320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932746355507487e-05,
+      "loss": 1.3538,
+      "step": 9330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9326701905646874e-05,
+      "loss": 1.426,
+      "step": 9340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932594025621887e-05,
+      "loss": 1.3944,
+      "step": 9350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932517860679087e-05,
+      "loss": 1.3992,
+      "step": 9360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932441695736287e-05,
+      "loss": 1.4766,
+      "step": 9370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932365530793487e-05,
+      "loss": 1.4809,
+      "step": 9380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932289365850686e-05,
+      "loss": 1.4189,
+      "step": 9390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9322132009078866e-05,
+      "loss": 1.3689,
+      "step": 9400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932137035965086e-05,
+      "loss": 1.464,
+      "step": 9410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.932060871022286e-05,
+      "loss": 1.5526,
+      "step": 9420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931984706079486e-05,
+      "loss": 1.3609,
+      "step": 9430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931908541136686e-05,
+      "loss": 1.3974,
+      "step": 9440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931832376193885e-05,
+      "loss": 1.4048,
+      "step": 9450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931756211251086e-05,
+      "loss": 1.418,
+      "step": 9460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9316800463082855e-05,
+      "loss": 1.3845,
+      "step": 9470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931603881365485e-05,
+      "loss": 1.4747,
+      "step": 9480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931527716422685e-05,
+      "loss": 1.3902,
+      "step": 9490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931451551479885e-05,
+      "loss": 1.464,
+      "step": 9500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931375386537085e-05,
+      "loss": 1.4311,
+      "step": 9510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.931299221594285e-05,
+      "loss": 1.2769,
+      "step": 9520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9312230566514847e-05,
+      "loss": 1.4834,
+      "step": 9530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9311468917086844e-05,
+      "loss": 1.258,
+      "step": 9540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.931070726765885e-05,
+      "loss": 1.2824,
+      "step": 9550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930994561823084e-05,
+      "loss": 1.5393,
+      "step": 9560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930918396880284e-05,
+      "loss": 1.3943,
+      "step": 9570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930842231937484e-05,
+      "loss": 1.4533,
+      "step": 9580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930766066994684e-05,
+      "loss": 1.2526,
+      "step": 9590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9306899020518836e-05,
+      "loss": 1.3994,
+      "step": 9600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930613737109084e-05,
+      "loss": 1.4273,
+      "step": 9610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930537572166283e-05,
+      "loss": 1.3983,
+      "step": 9620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9304614072234835e-05,
+      "loss": 1.3193,
+      "step": 9630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930385242280683e-05,
+      "loss": 1.4851,
+      "step": 9640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930309077337883e-05,
+      "loss": 1.5513,
+      "step": 9650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930232912395083e-05,
+      "loss": 1.3976,
+      "step": 9660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930156747452283e-05,
+      "loss": 1.4081,
+      "step": 9670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930080582509483e-05,
+      "loss": 1.3723,
+      "step": 9680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9300044175666826e-05,
+      "loss": 1.494,
+      "step": 9690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9299282526238824e-05,
+      "loss": 1.4724,
+      "step": 9700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.929852087681082e-05,
+      "loss": 1.246,
+      "step": 9710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9297759227382825e-05,
+      "loss": 1.4201,
+      "step": 9720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.929699757795482e-05,
+      "loss": 1.7487,
+      "step": 9730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.929623592852682e-05,
+      "loss": 1.2973,
+      "step": 9740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929547427909882e-05,
+      "loss": 1.4263,
+      "step": 9750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929471262967082e-05,
+      "loss": 1.4093,
+      "step": 9760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929395098024281e-05,
+      "loss": 1.3954,
+      "step": 9770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929318933081482e-05,
+      "loss": 1.3085,
+      "step": 9780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9292427681386815e-05,
+      "loss": 1.2673,
+      "step": 9790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929166603195881e-05,
+      "loss": 1.3415,
+      "step": 9800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929090438253081e-05,
+      "loss": 1.4845,
+      "step": 9810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9290142733102814e-05,
+      "loss": 1.3268,
+      "step": 9820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9289381083674804e-05,
+      "loss": 1.4184,
+      "step": 9830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.928861943424681e-05,
+      "loss": 1.4105,
+      "step": 9840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9287857784818806e-05,
+      "loss": 1.3718,
+      "step": 9850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9287096135390804e-05,
+      "loss": 1.4247,
+      "step": 9860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92863344859628e-05,
+      "loss": 1.4731,
+      "step": 9870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9285572836534805e-05,
+      "loss": 1.3728,
+      "step": 9880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92848111871068e-05,
+      "loss": 1.3626,
+      "step": 9890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92840495376788e-05,
+      "loss": 1.5367,
+      "step": 9900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92832878882508e-05,
+      "loss": 1.3951,
+      "step": 9910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9282526238822795e-05,
+      "loss": 1.3942,
+      "step": 9920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92817645893948e-05,
+      "loss": 1.3948,
+      "step": 9930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92810029399668e-05,
+      "loss": 1.4137,
+      "step": 9940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9280241290538794e-05,
+      "loss": 1.4198,
+      "step": 9950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.927947964111079e-05,
+      "loss": 1.3794,
+      "step": 9960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927871799168279e-05,
+      "loss": 1.4749,
+      "step": 9970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927795634225479e-05,
+      "loss": 1.588,
+      "step": 9980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927719469282679e-05,
+      "loss": 1.5276,
+      "step": 9990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927643304339878e-05,
+      "loss": 1.3214,
+      "step": 10000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9275671393970786e-05,
+      "loss": 1.4247,
+      "step": 10010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927490974454278e-05,
+      "loss": 1.3914,
+      "step": 10020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927414809511478e-05,
+      "loss": 1.4235,
+      "step": 10030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927338644568678e-05,
+      "loss": 1.3726,
+      "step": 10040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927262479625878e-05,
+      "loss": 1.4941,
+      "step": 10050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927186314683078e-05,
+      "loss": 1.2986,
+      "step": 10060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927110149740278e-05,
+      "loss": 1.456,
+      "step": 10070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9270339847974775e-05,
+      "loss": 1.3952,
+      "step": 10080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926957819854677e-05,
+      "loss": 1.3882,
+      "step": 10090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926881654911878e-05,
+      "loss": 1.4642,
+      "step": 10100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9268054899690774e-05,
+      "loss": 1.4007,
+      "step": 10110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926729325026277e-05,
+      "loss": 1.3334,
+      "step": 10120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926653160083477e-05,
+      "loss": 1.3946,
+      "step": 10130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926576995140677e-05,
+      "loss": 1.4604,
+      "step": 10140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9265008301978764e-05,
+      "loss": 1.5002,
+      "step": 10150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926424665255077e-05,
+      "loss": 1.3864,
+      "step": 10160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9263485003122766e-05,
+      "loss": 1.5012,
+      "step": 10170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926272335369476e-05,
+      "loss": 1.4132,
+      "step": 10180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.926196170426676e-05,
+      "loss": 1.402,
+      "step": 10190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9261200054838765e-05,
+      "loss": 1.6064,
+      "step": 10200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9260438405410756e-05,
+      "loss": 1.5362,
+      "step": 10210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925967675598276e-05,
+      "loss": 1.2405,
+      "step": 10220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925891510655476e-05,
+      "loss": 1.3935,
+      "step": 10230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9258153457126755e-05,
+      "loss": 1.408,
+      "step": 10240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925739180769875e-05,
+      "loss": 1.4143,
+      "step": 10250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9256630158270756e-05,
+      "loss": 1.2736,
+      "step": 10260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9255868508842754e-05,
+      "loss": 1.4537,
+      "step": 10270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925510685941475e-05,
+      "loss": 1.3211,
+      "step": 10280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925434520998675e-05,
+      "loss": 1.4506,
+      "step": 10290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9253583560558746e-05,
+      "loss": 1.4614,
+      "step": 10300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925282191113075e-05,
+      "loss": 1.4017,
+      "step": 10310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925206026170275e-05,
+      "loss": 1.3548,
+      "step": 10320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9251298612274745e-05,
+      "loss": 1.4713,
+      "step": 10330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.925053696284674e-05,
+      "loss": 1.4132,
+      "step": 10340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.924977531341875e-05,
+      "loss": 1.6246,
+      "step": 10350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.924901366399074e-05,
+      "loss": 1.402,
+      "step": 10360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.924825201456274e-05,
+      "loss": 1.449,
+      "step": 10370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.924749036513474e-05,
+      "loss": 1.4478,
+      "step": 10380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.924672871570674e-05,
+      "loss": 1.347,
+      "step": 10390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9245967066278735e-05,
+      "loss": 1.5242,
+      "step": 10400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.924520541685074e-05,
+      "loss": 1.4104,
+      "step": 10410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.924444376742273e-05,
+      "loss": 1.4647,
+      "step": 10420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9243682117994734e-05,
+      "loss": 1.4805,
+      "step": 10430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.924292046856673e-05,
+      "loss": 1.4007,
+      "step": 10440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.924215881913873e-05,
+      "loss": 1.3245,
+      "step": 10450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9241397169710726e-05,
+      "loss": 1.3285,
+      "step": 10460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9240635520282724e-05,
+      "loss": 1.4205,
+      "step": 10470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923987387085473e-05,
+      "loss": 1.3895,
+      "step": 10480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9239112221426725e-05,
+      "loss": 1.3618,
+      "step": 10490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923835057199872e-05,
+      "loss": 1.376,
+      "step": 10500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923758892257072e-05,
+      "loss": 1.4487,
+      "step": 10510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9236827273142724e-05,
+      "loss": 1.4089,
+      "step": 10520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9236065623714715e-05,
+      "loss": 1.5473,
+      "step": 10530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923530397428672e-05,
+      "loss": 1.4131,
+      "step": 10540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923454232485872e-05,
+      "loss": 1.5388,
+      "step": 10550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9233780675430714e-05,
+      "loss": 1.4018,
+      "step": 10560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923301902600271e-05,
+      "loss": 1.2757,
+      "step": 10570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9232257376574716e-05,
+      "loss": 1.5278,
+      "step": 10580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923149572714671e-05,
+      "loss": 1.5463,
+      "step": 10590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.923073407771871e-05,
+      "loss": 1.3626,
+      "step": 10600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.922997242829071e-05,
+      "loss": 1.288,
+      "step": 10610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9229210778862706e-05,
+      "loss": 1.2875,
+      "step": 10620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92284491294347e-05,
+      "loss": 1.4337,
+      "step": 10630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.922768748000671e-05,
+      "loss": 1.474,
+      "step": 10640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9226925830578705e-05,
+      "loss": 1.4067,
+      "step": 10650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92261641811507e-05,
+      "loss": 1.2876,
+      "step": 10660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92254025317227e-05,
+      "loss": 1.3163,
+      "step": 10670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92246408822947e-05,
+      "loss": 1.3331,
+      "step": 10680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92238792328667e-05,
+      "loss": 1.4362,
+      "step": 10690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92231175834387e-05,
+      "loss": 1.3527,
+      "step": 10700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92223559340107e-05,
+      "loss": 1.3096,
+      "step": 10710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9221594284582694e-05,
+      "loss": 1.5113,
+      "step": 10720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.92208326351547e-05,
+      "loss": 1.4901,
+      "step": 10730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.922007098572669e-05,
+      "loss": 1.2523,
+      "step": 10740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921930933629869e-05,
+      "loss": 1.42,
+      "step": 10750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921854768687069e-05,
+      "loss": 1.4121,
+      "step": 10760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921778603744269e-05,
+      "loss": 1.4016,
+      "step": 10770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9217024388014686e-05,
+      "loss": 1.4349,
+      "step": 10780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921626273858669e-05,
+      "loss": 1.4461,
+      "step": 10790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921550108915868e-05,
+      "loss": 1.4531,
+      "step": 10800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9214739439730685e-05,
+      "loss": 1.3416,
+      "step": 10810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921397779030268e-05,
+      "loss": 1.3384,
+      "step": 10820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921321614087468e-05,
+      "loss": 1.4796,
+      "step": 10830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.921245449144668e-05,
+      "loss": 1.2514,
+      "step": 10840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.921169284201868e-05,
+      "loss": 1.3293,
+      "step": 10850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.921093119259068e-05,
+      "loss": 1.5088,
+      "step": 10860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9210169543162676e-05,
+      "loss": 1.5198,
+      "step": 10870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9209407893734674e-05,
+      "loss": 1.3859,
+      "step": 10880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920864624430667e-05,
+      "loss": 1.319,
+      "step": 10890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9207884594878676e-05,
+      "loss": 1.3974,
+      "step": 10900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9207122945450666e-05,
+      "loss": 1.4996,
+      "step": 10910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920636129602267e-05,
+      "loss": 1.3755,
+      "step": 10920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920559964659467e-05,
+      "loss": 1.3746,
+      "step": 10930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9204837997166666e-05,
+      "loss": 1.2939,
+      "step": 10940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920407634773866e-05,
+      "loss": 1.377,
+      "step": 10950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920331469831067e-05,
+      "loss": 1.4885,
+      "step": 10960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920255304888266e-05,
+      "loss": 1.3947,
+      "step": 10970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920179139945466e-05,
+      "loss": 1.3737,
+      "step": 10980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920102975002666e-05,
+      "loss": 1.4851,
+      "step": 10990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.920026810059866e-05,
+      "loss": 1.4191,
+      "step": 11000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9199506451170655e-05,
+      "loss": 1.558,
+      "step": 11010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.919874480174266e-05,
+      "loss": 1.5036,
+      "step": 11020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.919798315231465e-05,
+      "loss": 1.5179,
+      "step": 11030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9197221502886654e-05,
+      "loss": 1.2831,
+      "step": 11040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.919645985345865e-05,
+      "loss": 1.3723,
+      "step": 11050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919569820403065e-05,
+      "loss": 1.5146,
+      "step": 11060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919493655460265e-05,
+      "loss": 1.3656,
+      "step": 11070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919417490517465e-05,
+      "loss": 1.3468,
+      "step": 11080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919341325574665e-05,
+      "loss": 1.5983,
+      "step": 11090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9192651606318645e-05,
+      "loss": 1.5452,
+      "step": 11100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919188995689065e-05,
+      "loss": 1.4933,
+      "step": 11110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.919112830746264e-05,
+      "loss": 1.2555,
+      "step": 11120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9190366658034644e-05,
+      "loss": 1.4077,
+      "step": 11130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918960500860664e-05,
+      "loss": 1.3483,
+      "step": 11140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918884335917864e-05,
+      "loss": 1.3617,
+      "step": 11150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918808170975064e-05,
+      "loss": 1.4188,
+      "step": 11160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918732006032264e-05,
+      "loss": 1.3169,
+      "step": 11170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918655841089463e-05,
+      "loss": 1.4848,
+      "step": 11180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9185796761466636e-05,
+      "loss": 1.3899,
+      "step": 11190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9185035112038634e-05,
+      "loss": 1.2976,
+      "step": 11200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918427346261063e-05,
+      "loss": 1.3403,
+      "step": 11210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918351181318263e-05,
+      "loss": 1.3406,
+      "step": 11220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918275016375463e-05,
+      "loss": 1.4466,
+      "step": 11230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918198851432663e-05,
+      "loss": 1.3875,
+      "step": 11240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918122686489863e-05,
+      "loss": 1.3595,
+      "step": 11250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9180465215470625e-05,
+      "loss": 1.2402,
+      "step": 11260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.917970356604262e-05,
+      "loss": 1.4128,
+      "step": 11270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917894191661463e-05,
+      "loss": 1.4164,
+      "step": 11280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9178180267186624e-05,
+      "loss": 1.411,
+      "step": 11290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917741861775862e-05,
+      "loss": 1.2748,
+      "step": 11300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917665696833062e-05,
+      "loss": 1.3147,
+      "step": 11310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9175895318902623e-05,
+      "loss": 1.2332,
+      "step": 11320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9175133669474614e-05,
+      "loss": 1.4099,
+      "step": 11330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917437202004662e-05,
+      "loss": 1.4187,
+      "step": 11340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917361037061861e-05,
+      "loss": 1.3448,
+      "step": 11350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917284872119061e-05,
+      "loss": 1.4279,
+      "step": 11360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917208707176261e-05,
+      "loss": 1.3432,
+      "step": 11370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917132542233461e-05,
+      "loss": 1.3969,
+      "step": 11380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9170563772906606e-05,
+      "loss": 1.4577,
+      "step": 11390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916980212347861e-05,
+      "loss": 1.3351,
+      "step": 11400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91690404740506e-05,
+      "loss": 1.4166,
+      "step": 11410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9168278824622605e-05,
+      "loss": 1.3656,
+      "step": 11420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91675171751946e-05,
+      "loss": 1.2944,
+      "step": 11430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91667555257666e-05,
+      "loss": 1.4017,
+      "step": 11440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9165993876338604e-05,
+      "loss": 1.3942,
+      "step": 11450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91652322269106e-05,
+      "loss": 1.4572,
+      "step": 11460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91644705774826e-05,
+      "loss": 1.3886,
+      "step": 11470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9163708928054596e-05,
+      "loss": 1.4644,
+      "step": 11480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91629472786266e-05,
+      "loss": 1.4868,
+      "step": 11490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.916218562919859e-05,
+      "loss": 1.4188,
+      "step": 11500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9161423979770596e-05,
+      "loss": 1.2867,
+      "step": 11510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.916066233034259e-05,
+      "loss": 1.3943,
+      "step": 11520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915990068091459e-05,
+      "loss": 1.3641,
+      "step": 11530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915913903148659e-05,
+      "loss": 1.4144,
+      "step": 11540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915837738205859e-05,
+      "loss": 1.462,
+      "step": 11550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915761573263058e-05,
+      "loss": 1.2742,
+      "step": 11560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915685408320259e-05,
+      "loss": 1.4384,
+      "step": 11570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9156092433774585e-05,
+      "loss": 1.2891,
+      "step": 11580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915533078434658e-05,
+      "loss": 1.3795,
+      "step": 11590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915456913491858e-05,
+      "loss": 1.2751,
+      "step": 11600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9153807485490584e-05,
+      "loss": 1.2837,
+      "step": 11610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9153045836062575e-05,
+      "loss": 1.3611,
+      "step": 11620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.915228418663458e-05,
+      "loss": 1.401,
+      "step": 11630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9151522537206576e-05,
+      "loss": 1.3838,
+      "step": 11640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9150760887778574e-05,
+      "loss": 1.4422,
+      "step": 11650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.914999923835058e-05,
+      "loss": 1.4876,
+      "step": 11660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9149237588922575e-05,
+      "loss": 1.3878,
+      "step": 11670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.914847593949457e-05,
+      "loss": 1.3885,
+      "step": 11680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.914771429006657e-05,
+      "loss": 1.3443,
+      "step": 11690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9146952640638575e-05,
+      "loss": 1.4463,
+      "step": 11700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9146190991210565e-05,
+      "loss": 1.656,
+      "step": 11710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914542934178257e-05,
+      "loss": 1.3896,
+      "step": 11720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914466769235457e-05,
+      "loss": 1.4673,
+      "step": 11730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9143906042926564e-05,
+      "loss": 1.3279,
+      "step": 11740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914314439349856e-05,
+      "loss": 1.1601,
+      "step": 11750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9142382744070566e-05,
+      "loss": 1.2964,
+      "step": 11760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914162109464256e-05,
+      "loss": 1.2295,
+      "step": 11770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914085944521456e-05,
+      "loss": 1.4477,
+      "step": 11780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914009779578656e-05,
+      "loss": 1.4311,
+      "step": 11790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9139336146358556e-05,
+      "loss": 1.4121,
+      "step": 11800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9138574496930554e-05,
+      "loss": 1.3487,
+      "step": 11810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913781284750255e-05,
+      "loss": 1.3331,
+      "step": 11820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913705119807455e-05,
+      "loss": 1.2478,
+      "step": 11830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913628954864655e-05,
+      "loss": 1.4175,
+      "step": 11840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913552789921855e-05,
+      "loss": 1.2884,
+      "step": 11850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913476624979055e-05,
+      "loss": 1.3614,
+      "step": 11860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913400460036255e-05,
+      "loss": 1.4651,
+      "step": 11870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913324295093454e-05,
+      "loss": 1.2719,
+      "step": 11880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913248130150655e-05,
+      "loss": 1.4233,
+      "step": 11890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9131719652078544e-05,
+      "loss": 1.2185,
+      "step": 11900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913095800265054e-05,
+      "loss": 1.4053,
+      "step": 11910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913019635322254e-05,
+      "loss": 1.315,
+      "step": 11920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9129434703794543e-05,
+      "loss": 1.2746,
+      "step": 11930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9128673054366534e-05,
+      "loss": 1.5315,
+      "step": 11940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912791140493854e-05,
+      "loss": 1.4588,
+      "step": 11950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9127149755510536e-05,
+      "loss": 1.3813,
+      "step": 11960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912638810608253e-05,
+      "loss": 1.2318,
+      "step": 11970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912562645665453e-05,
+      "loss": 1.2305,
+      "step": 11980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9124864807226535e-05,
+      "loss": 1.5753,
+      "step": 11990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9124103157798526e-05,
+      "loss": 1.4071,
+      "step": 12000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912334150837053e-05,
+      "loss": 1.293,
+      "step": 12010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912257985894253e-05,
+      "loss": 1.4685,
+      "step": 12020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9121818209514525e-05,
+      "loss": 1.3544,
+      "step": 12030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912105656008653e-05,
+      "loss": 1.316,
+      "step": 12040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.912029491065853e-05,
+      "loss": 1.3141,
+      "step": 12050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9119533261230524e-05,
+      "loss": 1.5827,
+      "step": 12060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.911877161180252e-05,
+      "loss": 1.5074,
+      "step": 12070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9118009962374526e-05,
+      "loss": 1.3924,
+      "step": 12080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9117248312946516e-05,
+      "loss": 1.5574,
+      "step": 12090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.911648666351852e-05,
+      "loss": 1.2335,
+      "step": 12100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.911572501409052e-05,
+      "loss": 1.3286,
+      "step": 12110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9114963364662516e-05,
+      "loss": 1.3664,
+      "step": 12120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.911420171523451e-05,
+      "loss": 1.414,
+      "step": 12130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.911344006580652e-05,
+      "loss": 1.3114,
+      "step": 12140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.911267841637851e-05,
+      "loss": 1.3488,
+      "step": 12150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911191676695051e-05,
+      "loss": 1.3081,
+      "step": 12160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911115511752251e-05,
+      "loss": 1.4271,
+      "step": 12170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911039346809451e-05,
+      "loss": 1.3449,
+      "step": 12180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9109631818666505e-05,
+      "loss": 1.4907,
+      "step": 12190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910887016923851e-05,
+      "loss": 1.3187,
+      "step": 12200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91081085198105e-05,
+      "loss": 1.3351,
+      "step": 12210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9107346870382504e-05,
+      "loss": 1.4149,
+      "step": 12220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91065852209545e-05,
+      "loss": 1.3958,
+      "step": 12230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91058235715265e-05,
+      "loss": 1.3486,
+      "step": 12240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91050619220985e-05,
+      "loss": 1.409,
+      "step": 12250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9104300272670494e-05,
+      "loss": 1.4862,
+      "step": 12260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91035386232425e-05,
+      "loss": 1.2595,
+      "step": 12270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9102776973814495e-05,
+      "loss": 1.3843,
+      "step": 12280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910201532438649e-05,
+      "loss": 1.2102,
+      "step": 12290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910125367495849e-05,
+      "loss": 1.3443,
+      "step": 12300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9100492025530495e-05,
+      "loss": 1.4276,
+      "step": 12310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9099730376102485e-05,
+      "loss": 1.4234,
+      "step": 12320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.909896872667449e-05,
+      "loss": 1.3169,
+      "step": 12330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.909820707724649e-05,
+      "loss": 1.3885,
+      "step": 12340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9097445427818484e-05,
+      "loss": 1.4717,
+      "step": 12350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.909668377839048e-05,
+      "loss": 1.3374,
+      "step": 12360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9095922128962486e-05,
+      "loss": 1.4435,
+      "step": 12370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909516047953448e-05,
+      "loss": 1.4675,
+      "step": 12380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909439883010648e-05,
+      "loss": 1.3238,
+      "step": 12390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909363718067848e-05,
+      "loss": 1.3932,
+      "step": 12400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9092875531250476e-05,
+      "loss": 1.5465,
+      "step": 12410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9092113881822474e-05,
+      "loss": 1.3324,
+      "step": 12420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909135223239448e-05,
+      "loss": 1.4936,
+      "step": 12430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9090590582966475e-05,
+      "loss": 1.2713,
+      "step": 12440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908982893353847e-05,
+      "loss": 1.2698,
+      "step": 12450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908906728411048e-05,
+      "loss": 1.4365,
+      "step": 12460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908830563468247e-05,
+      "loss": 1.3584,
+      "step": 12470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908754398525447e-05,
+      "loss": 1.24,
+      "step": 12480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908678233582647e-05,
+      "loss": 1.4839,
+      "step": 12490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908602068639847e-05,
+      "loss": 1.4633,
+      "step": 12500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9085259036970464e-05,
+      "loss": 1.4854,
+      "step": 12510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908449738754247e-05,
+      "loss": 1.35,
+      "step": 12520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908373573811446e-05,
+      "loss": 1.5274,
+      "step": 12530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9082974088686463e-05,
+      "loss": 1.3316,
+      "step": 12540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908221243925846e-05,
+      "loss": 1.4132,
+      "step": 12550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908145078983046e-05,
+      "loss": 1.2259,
+      "step": 12560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9080689140402456e-05,
+      "loss": 1.3263,
+      "step": 12570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907992749097446e-05,
+      "loss": 1.3995,
+      "step": 12580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907916584154645e-05,
+      "loss": 1.3517,
+      "step": 12590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9078404192118455e-05,
+      "loss": 1.5033,
+      "step": 12600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907764254269045e-05,
+      "loss": 1.3562,
+      "step": 12610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907688089326245e-05,
+      "loss": 1.5306,
+      "step": 12620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907611924383445e-05,
+      "loss": 1.3127,
+      "step": 12630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907535759440645e-05,
+      "loss": 1.2992,
+      "step": 12640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907459594497845e-05,
+      "loss": 1.3442,
+      "step": 12650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907383429555045e-05,
+      "loss": 1.164,
+      "step": 12660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907307264612245e-05,
+      "loss": 1.3616,
+      "step": 12670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907231099669444e-05,
+      "loss": 1.3583,
+      "step": 12680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9071549347266446e-05,
+      "loss": 1.3225,
+      "step": 12690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907078769783844e-05,
+      "loss": 1.5161,
+      "step": 12700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.907002604841044e-05,
+      "loss": 1.3338,
+      "step": 12710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906926439898244e-05,
+      "loss": 1.4392,
+      "step": 12720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9068502749554436e-05,
+      "loss": 1.3852,
+      "step": 12730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906774110012643e-05,
+      "loss": 1.2123,
+      "step": 12740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906697945069844e-05,
+      "loss": 1.3297,
+      "step": 12750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906621780127043e-05,
+      "loss": 1.22,
+      "step": 12760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906545615184243e-05,
+      "loss": 1.3669,
+      "step": 12770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906469450241443e-05,
+      "loss": 1.2326,
+      "step": 12780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906393285298643e-05,
+      "loss": 1.3881,
+      "step": 12790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9063171203558425e-05,
+      "loss": 1.2143,
+      "step": 12800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.906240955413043e-05,
+      "loss": 1.5362,
+      "step": 12810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9061647904702426e-05,
+      "loss": 1.4099,
+      "step": 12820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9060886255274424e-05,
+      "loss": 1.3789,
+      "step": 12830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.906012460584643e-05,
+      "loss": 1.3282,
+      "step": 12840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905936295641842e-05,
+      "loss": 1.3462,
+      "step": 12850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905860130699042e-05,
+      "loss": 1.3798,
+      "step": 12860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905783965756242e-05,
+      "loss": 1.1915,
+      "step": 12870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905707800813442e-05,
+      "loss": 1.4207,
+      "step": 12880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9056316358706415e-05,
+      "loss": 1.4254,
+      "step": 12890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905555470927842e-05,
+      "loss": 1.582,
+      "step": 12900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905479305985041e-05,
+      "loss": 1.4282,
+      "step": 12910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9054031410422415e-05,
+      "loss": 1.3641,
+      "step": 12920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905326976099441e-05,
+      "loss": 1.4693,
+      "step": 12930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905250811156641e-05,
+      "loss": 1.4313,
+      "step": 12940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905174646213841e-05,
+      "loss": 1.4903,
+      "step": 12950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.905098481271041e-05,
+      "loss": 1.4583,
+      "step": 12960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90502231632824e-05,
+      "loss": 1.302,
+      "step": 12970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9049461513854406e-05,
+      "loss": 1.3123,
+      "step": 12980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9048699864426404e-05,
+      "loss": 1.3635,
+      "step": 12990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90479382149984e-05,
+      "loss": 1.3541,
+      "step": 13000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90471765655704e-05,
+      "loss": 1.4922,
+      "step": 13010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90464149161424e-05,
+      "loss": 1.3202,
+      "step": 13020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90456532667144e-05,
+      "loss": 1.4314,
+      "step": 13030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.90448916172864e-05,
+      "loss": 1.4767,
+      "step": 13040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.90441299678584e-05,
+      "loss": 1.4686,
+      "step": 13050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.904336831843039e-05,
+      "loss": 1.2333,
+      "step": 13060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.90426066690024e-05,
+      "loss": 1.2404,
+      "step": 13070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9041845019574394e-05,
+      "loss": 1.3516,
+      "step": 13080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.904108337014639e-05,
+      "loss": 1.4431,
+      "step": 13090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.904032172071839e-05,
+      "loss": 1.3424,
+      "step": 13100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9039560071290394e-05,
+      "loss": 1.3072,
+      "step": 13110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9038798421862384e-05,
+      "loss": 1.3214,
+      "step": 13120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.903803677243439e-05,
+      "loss": 1.2547,
+      "step": 13130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9037275123006386e-05,
+      "loss": 1.3391,
+      "step": 13140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9036513473578383e-05,
+      "loss": 1.463,
+      "step": 13150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.903575182415038e-05,
+      "loss": 1.3157,
+      "step": 13160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9034990174722385e-05,
+      "loss": 1.3504,
+      "step": 13170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9034228525294376e-05,
+      "loss": 1.2397,
+      "step": 13180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.903346687586638e-05,
+      "loss": 1.3398,
+      "step": 13190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.903270522643838e-05,
+      "loss": 1.257,
+      "step": 13200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9031943577010375e-05,
+      "loss": 1.3279,
+      "step": 13210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.903118192758237e-05,
+      "loss": 1.2328,
+      "step": 13220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.903042027815437e-05,
+      "loss": 1.3049,
+      "step": 13230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9029658628726374e-05,
+      "loss": 1.4208,
+      "step": 13240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902889697929837e-05,
+      "loss": 1.3165,
+      "step": 13250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902813532987037e-05,
+      "loss": 1.3036,
+      "step": 13260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902737368044237e-05,
+      "loss": 1.4194,
+      "step": 13270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902661203101437e-05,
+      "loss": 1.2832,
+      "step": 13280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902585038158636e-05,
+      "loss": 1.3585,
+      "step": 13290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9025088732158366e-05,
+      "loss": 1.3415,
+      "step": 13300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902432708273036e-05,
+      "loss": 1.3748,
+      "step": 13310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902356543330236e-05,
+      "loss": 1.2032,
+      "step": 13320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902280378387436e-05,
+      "loss": 1.3026,
+      "step": 13330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902204213444636e-05,
+      "loss": 1.378,
+      "step": 13340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902128048501835e-05,
+      "loss": 1.4106,
+      "step": 13350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.902051883559036e-05,
+      "loss": 1.2801,
+      "step": 13360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9019757186162355e-05,
+      "loss": 1.3379,
+      "step": 13370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.901899553673435e-05,
+      "loss": 1.4603,
+      "step": 13380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.901823388730635e-05,
+      "loss": 1.3746,
+      "step": 13390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9017472237878354e-05,
+      "loss": 1.2747,
+      "step": 13400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.901671058845035e-05,
+      "loss": 1.4365,
+      "step": 13410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.901594893902235e-05,
+      "loss": 1.4241,
+      "step": 13420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9015187289594346e-05,
+      "loss": 1.2279,
+      "step": 13430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9014425640166344e-05,
+      "loss": 1.4234,
+      "step": 13440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.901366399073835e-05,
+      "loss": 1.3044,
+      "step": 13450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9012902341310346e-05,
+      "loss": 1.3363,
+      "step": 13460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.901214069188234e-05,
+      "loss": 1.4935,
+      "step": 13470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.901137904245434e-05,
+      "loss": 1.2657,
+      "step": 13480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9010617393026345e-05,
+      "loss": 1.2798,
+      "step": 13490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9009855743598335e-05,
+      "loss": 1.5163,
+      "step": 13500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900909409417034e-05,
+      "loss": 1.4239,
+      "step": 13510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900833244474234e-05,
+      "loss": 1.3034,
+      "step": 13520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9007570795314335e-05,
+      "loss": 1.5107,
+      "step": 13530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900680914588633e-05,
+      "loss": 1.3206,
+      "step": 13540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9006047496458336e-05,
+      "loss": 1.269,
+      "step": 13550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900528584703033e-05,
+      "loss": 1.4041,
+      "step": 13560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900452419760233e-05,
+      "loss": 1.3229,
+      "step": 13570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900376254817433e-05,
+      "loss": 1.2939,
+      "step": 13580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9003000898746326e-05,
+      "loss": 1.2381,
+      "step": 13590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9002239249318324e-05,
+      "loss": 1.2457,
+      "step": 13600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900147759989033e-05,
+      "loss": 1.4375,
+      "step": 13610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9000715950462325e-05,
+      "loss": 1.3648,
+      "step": 13620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.899995430103432e-05,
+      "loss": 1.2406,
+      "step": 13630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.899919265160632e-05,
+      "loss": 1.4574,
+      "step": 13640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.899843100217832e-05,
+      "loss": 1.3764,
+      "step": 13650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.899766935275032e-05,
+      "loss": 1.3497,
+      "step": 13660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.899690770332231e-05,
+      "loss": 1.2489,
+      "step": 13670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.899614605389432e-05,
+      "loss": 1.3669,
+      "step": 13680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8995384404466314e-05,
+      "loss": 1.2022,
+      "step": 13690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.899462275503831e-05,
+      "loss": 1.2263,
+      "step": 13700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.899386110561031e-05,
+      "loss": 1.4551,
+      "step": 13710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8993099456182314e-05,
+      "loss": 1.2936,
+      "step": 13720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8992337806754304e-05,
+      "loss": 1.3626,
+      "step": 13730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.899157615732631e-05,
+      "loss": 1.3546,
+      "step": 13740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8990814507898306e-05,
+      "loss": 1.3413,
+      "step": 13750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8990052858470303e-05,
+      "loss": 1.3947,
+      "step": 13760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89892912090423e-05,
+      "loss": 1.3867,
+      "step": 13770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8988529559614305e-05,
+      "loss": 1.222,
+      "step": 13780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89877679101863e-05,
+      "loss": 1.3849,
+      "step": 13790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89870062607583e-05,
+      "loss": 1.2832,
+      "step": 13800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89862446113303e-05,
+      "loss": 1.3428,
+      "step": 13810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8985482961902295e-05,
+      "loss": 1.2705,
+      "step": 13820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89847213124743e-05,
+      "loss": 1.5269,
+      "step": 13830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89839596630463e-05,
+      "loss": 1.3868,
+      "step": 13840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8983198013618294e-05,
+      "loss": 1.391,
+      "step": 13850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.898243636419029e-05,
+      "loss": 1.2979,
+      "step": 13860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8981674714762296e-05,
+      "loss": 1.3796,
+      "step": 13870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.898091306533429e-05,
+      "loss": 1.2472,
+      "step": 13880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.898015141590629e-05,
+      "loss": 1.2427,
+      "step": 13890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.897938976647829e-05,
+      "loss": 1.3,
+      "step": 13900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8978628117050286e-05,
+      "loss": 1.3381,
+      "step": 13910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897786646762228e-05,
+      "loss": 1.3384,
+      "step": 13920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897710481819429e-05,
+      "loss": 1.4341,
+      "step": 13930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897634316876628e-05,
+      "loss": 1.302,
+      "step": 13940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897558151933828e-05,
+      "loss": 1.4718,
+      "step": 13950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897481986991028e-05,
+      "loss": 1.4117,
+      "step": 13960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897405822048228e-05,
+      "loss": 1.3141,
+      "step": 13970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8973296571054275e-05,
+      "loss": 1.1666,
+      "step": 13980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897253492162628e-05,
+      "loss": 1.4563,
+      "step": 13990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8971773272198277e-05,
+      "loss": 1.1705,
+      "step": 14000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8971011622770274e-05,
+      "loss": 1.4487,
+      "step": 14010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.897024997334227e-05,
+      "loss": 1.344,
+      "step": 14020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896948832391427e-05,
+      "loss": 1.4043,
+      "step": 14030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896872667448627e-05,
+      "loss": 1.2829,
+      "step": 14040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896796502505827e-05,
+      "loss": 1.1871,
+      "step": 14050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896720337563027e-05,
+      "loss": 1.2817,
+      "step": 14060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8966441726202266e-05,
+      "loss": 1.2453,
+      "step": 14070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896568007677427e-05,
+      "loss": 1.4368,
+      "step": 14080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896491842734626e-05,
+      "loss": 1.3099,
+      "step": 14090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8964156777918265e-05,
+      "loss": 1.4106,
+      "step": 14100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8963395128490255e-05,
+      "loss": 1.4937,
+      "step": 14110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.896263347906226e-05,
+      "loss": 1.3771,
+      "step": 14120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.896187182963426e-05,
+      "loss": 1.4332,
+      "step": 14130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8961110180206255e-05,
+      "loss": 1.3459,
+      "step": 14140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.896034853077825e-05,
+      "loss": 1.3187,
+      "step": 14150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8959586881350256e-05,
+      "loss": 1.4126,
+      "step": 14160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8958825231922254e-05,
+      "loss": 1.4581,
+      "step": 14170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895806358249425e-05,
+      "loss": 1.3624,
+      "step": 14180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895730193306625e-05,
+      "loss": 1.2522,
+      "step": 14190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8956540283638246e-05,
+      "loss": 1.42,
+      "step": 14200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895577863421025e-05,
+      "loss": 1.4283,
+      "step": 14210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895501698478225e-05,
+      "loss": 1.1841,
+      "step": 14220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8954255335354245e-05,
+      "loss": 1.3737,
+      "step": 14230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895349368592624e-05,
+      "loss": 1.2149,
+      "step": 14240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895273203649825e-05,
+      "loss": 1.299,
+      "step": 14250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895197038707024e-05,
+      "loss": 1.2421,
+      "step": 14260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895120873764224e-05,
+      "loss": 1.1488,
+      "step": 14270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.895044708821424e-05,
+      "loss": 1.2446,
+      "step": 14280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.894968543878624e-05,
+      "loss": 1.2389,
+      "step": 14290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8948923789358234e-05,
+      "loss": 1.1944,
+      "step": 14300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.894816213993024e-05,
+      "loss": 1.3149,
+      "step": 14310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.894740049050223e-05,
+      "loss": 1.3809,
+      "step": 14320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8946638841074234e-05,
+      "loss": 1.2641,
+      "step": 14330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.894587719164623e-05,
+      "loss": 1.3284,
+      "step": 14340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.894511554221823e-05,
+      "loss": 1.3529,
+      "step": 14350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8944353892790226e-05,
+      "loss": 1.3423,
+      "step": 14360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.894359224336223e-05,
+      "loss": 1.3795,
+      "step": 14370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.894283059393423e-05,
+      "loss": 1.3379,
+      "step": 14380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8942068944506225e-05,
+      "loss": 1.3713,
+      "step": 14390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.894130729507822e-05,
+      "loss": 1.2344,
+      "step": 14400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.894054564565022e-05,
+      "loss": 1.3494,
+      "step": 14410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8939783996222224e-05,
+      "loss": 1.2778,
+      "step": 14420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893902234679422e-05,
+      "loss": 1.4684,
+      "step": 14430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893826069736622e-05,
+      "loss": 1.2985,
+      "step": 14440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893749904793822e-05,
+      "loss": 1.3241,
+      "step": 14450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893673739851022e-05,
+      "loss": 1.3371,
+      "step": 14460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893597574908221e-05,
+      "loss": 1.2777,
+      "step": 14470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8935214099654216e-05,
+      "loss": 1.2949,
+      "step": 14480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8934452450226213e-05,
+      "loss": 1.5325,
+      "step": 14490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893369080079821e-05,
+      "loss": 1.4687,
+      "step": 14500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893292915137021e-05,
+      "loss": 1.2447,
+      "step": 14510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893216750194221e-05,
+      "loss": 1.3857,
+      "step": 14520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.89314058525142e-05,
+      "loss": 1.141,
+      "step": 14530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.893064420308621e-05,
+      "loss": 1.4047,
+      "step": 14540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8929882553658205e-05,
+      "loss": 1.2231,
+      "step": 14550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.89291209042302e-05,
+      "loss": 1.2752,
+      "step": 14560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89283592548022e-05,
+      "loss": 1.2672,
+      "step": 14570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89275976053742e-05,
+      "loss": 1.3777,
+      "step": 14580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89268359559462e-05,
+      "loss": 1.2886,
+      "step": 14590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89260743065182e-05,
+      "loss": 1.2908,
+      "step": 14600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8925312657090197e-05,
+      "loss": 1.2169,
+      "step": 14610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8924551007662194e-05,
+      "loss": 1.412,
+      "step": 14620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89237893582342e-05,
+      "loss": 1.4146,
+      "step": 14630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.892302770880619e-05,
+      "loss": 1.4703,
+      "step": 14640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.892226605937819e-05,
+      "loss": 1.3097,
+      "step": 14650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.892150440995019e-05,
+      "loss": 1.19,
+      "step": 14660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.892074276052219e-05,
+      "loss": 1.417,
+      "step": 14670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8919981111094186e-05,
+      "loss": 1.3816,
+      "step": 14680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891921946166619e-05,
+      "loss": 1.3117,
+      "step": 14690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891845781223818e-05,
+      "loss": 1.3573,
+      "step": 14700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8917696162810185e-05,
+      "loss": 1.3932,
+      "step": 14710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891693451338218e-05,
+      "loss": 1.3065,
+      "step": 14720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891617286395418e-05,
+      "loss": 1.2899,
+      "step": 14730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891541121452618e-05,
+      "loss": 1.369,
+      "step": 14740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891464956509818e-05,
+      "loss": 1.1461,
+      "step": 14750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891388791567018e-05,
+      "loss": 1.2857,
+      "step": 14760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8913126266242176e-05,
+      "loss": 1.2572,
+      "step": 14770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8912364616814174e-05,
+      "loss": 1.2453,
+      "step": 14780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891160296738617e-05,
+      "loss": 1.519,
+      "step": 14790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8910841317958176e-05,
+      "loss": 1.1898,
+      "step": 14800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891007966853017e-05,
+      "loss": 1.2981,
+      "step": 14810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890931801910217e-05,
+      "loss": 1.317,
+      "step": 14820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890855636967417e-05,
+      "loss": 1.2265,
+      "step": 14830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890779472024617e-05,
+      "loss": 1.357,
+      "step": 14840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890703307081816e-05,
+      "loss": 1.332,
+      "step": 14850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890627142139017e-05,
+      "loss": 1.3536,
+      "step": 14860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8905509771962165e-05,
+      "loss": 1.3805,
+      "step": 14870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890474812253416e-05,
+      "loss": 1.1781,
+      "step": 14880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890398647310616e-05,
+      "loss": 1.3168,
+      "step": 14890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8903224823678164e-05,
+      "loss": 1.3853,
+      "step": 14900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8902463174250154e-05,
+      "loss": 1.3786,
+      "step": 14910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890170152482216e-05,
+      "loss": 1.3965,
+      "step": 14920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8900939875394156e-05,
+      "loss": 1.1962,
+      "step": 14930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8900178225966154e-05,
+      "loss": 1.2863,
+      "step": 14940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889941657653815e-05,
+      "loss": 1.3693,
+      "step": 14950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8898654927110155e-05,
+      "loss": 1.1671,
+      "step": 14960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889789327768215e-05,
+      "loss": 1.1418,
+      "step": 14970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889713162825415e-05,
+      "loss": 1.2834,
+      "step": 14980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889636997882615e-05,
+      "loss": 1.3773,
+      "step": 14990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8895608329398145e-05,
+      "loss": 1.2862,
+      "step": 15000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.889484667997015e-05,
+      "loss": 1.4312,
+      "step": 15010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.889408503054215e-05,
+      "loss": 1.182,
+      "step": 15020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8893323381114144e-05,
+      "loss": 1.4415,
+      "step": 15030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.889256173168614e-05,
+      "loss": 1.3276,
+      "step": 15040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.889180008225814e-05,
+      "loss": 1.2088,
+      "step": 15050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.889103843283014e-05,
+      "loss": 1.2974,
+      "step": 15060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.889027678340214e-05,
+      "loss": 1.4854,
+      "step": 15070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888951513397413e-05,
+      "loss": 1.3259,
+      "step": 15080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8888753484546136e-05,
+      "loss": 1.3408,
+      "step": 15090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8887991835118133e-05,
+      "loss": 1.3268,
+      "step": 15100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888723018569013e-05,
+      "loss": 1.3557,
+      "step": 15110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888646853626213e-05,
+      "loss": 1.3296,
+      "step": 15120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888570688683413e-05,
+      "loss": 1.3889,
+      "step": 15130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888494523740613e-05,
+      "loss": 1.3135,
+      "step": 15140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888418358797813e-05,
+      "loss": 1.2622,
+      "step": 15150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8883421938550125e-05,
+      "loss": 1.3913,
+      "step": 15160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888266028912212e-05,
+      "loss": 1.3468,
+      "step": 15170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888189863969413e-05,
+      "loss": 1.228,
+      "step": 15180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8881136990266124e-05,
+      "loss": 1.2783,
+      "step": 15190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.888037534083812e-05,
+      "loss": 1.3969,
+      "step": 15200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.887961369141012e-05,
+      "loss": 1.3892,
+      "step": 15210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.887885204198212e-05,
+      "loss": 1.2734,
+      "step": 15220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8878090392554114e-05,
+      "loss": 1.2757,
+      "step": 15230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.887732874312612e-05,
+      "loss": 1.3062,
+      "step": 15240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8876567093698116e-05,
+      "loss": 1.321,
+      "step": 15250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.887580544427011e-05,
+      "loss": 1.381,
+      "step": 15260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.887504379484211e-05,
+      "loss": 1.2757,
+      "step": 15270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8874282145414115e-05,
+      "loss": 1.5321,
+      "step": 15280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8873520495986106e-05,
+      "loss": 1.2218,
+      "step": 15290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.887275884655811e-05,
+      "loss": 1.1998,
+      "step": 15300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.887199719713011e-05,
+      "loss": 1.1642,
+      "step": 15310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8871235547702105e-05,
+      "loss": 1.3518,
+      "step": 15320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88704738982741e-05,
+      "loss": 1.2855,
+      "step": 15330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8869712248846107e-05,
+      "loss": 1.2074,
+      "step": 15340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8868950599418104e-05,
+      "loss": 1.4778,
+      "step": 15350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88681889499901e-05,
+      "loss": 1.36,
+      "step": 15360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88674273005621e-05,
+      "loss": 1.4415,
+      "step": 15370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8866665651134096e-05,
+      "loss": 1.2429,
+      "step": 15380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88659040017061e-05,
+      "loss": 1.3725,
+      "step": 15390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88651423522781e-05,
+      "loss": 1.353,
+      "step": 15400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8864380702850096e-05,
+      "loss": 1.4396,
+      "step": 15410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.886361905342209e-05,
+      "loss": 1.376,
+      "step": 15420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88628574039941e-05,
+      "loss": 1.3347,
+      "step": 15430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.886209575456609e-05,
+      "loss": 1.3009,
+      "step": 15440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.886133410513809e-05,
+      "loss": 1.2814,
+      "step": 15450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.886057245571009e-05,
+      "loss": 1.2497,
+      "step": 15460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885981080628209e-05,
+      "loss": 1.3844,
+      "step": 15470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8859049156854085e-05,
+      "loss": 1.3046,
+      "step": 15480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885828750742608e-05,
+      "loss": 1.4392,
+      "step": 15490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885752585799808e-05,
+      "loss": 1.1335,
+      "step": 15500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8856764208570084e-05,
+      "loss": 1.3052,
+      "step": 15510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8856002559142074e-05,
+      "loss": 1.2963,
+      "step": 15520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885524090971408e-05,
+      "loss": 1.2648,
+      "step": 15530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8854479260286076e-05,
+      "loss": 1.2448,
+      "step": 15540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8853717610858074e-05,
+      "loss": 1.3127,
+      "step": 15550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885295596143008e-05,
+      "loss": 1.4304,
+      "step": 15560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8852194312002075e-05,
+      "loss": 1.2862,
+      "step": 15570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885143266257407e-05,
+      "loss": 1.3307,
+      "step": 15580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.885067101314607e-05,
+      "loss": 1.1718,
+      "step": 15590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8849909363718075e-05,
+      "loss": 1.3437,
+      "step": 15600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8849147714290065e-05,
+      "loss": 1.369,
+      "step": 15610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.884838606486207e-05,
+      "loss": 1.2092,
+      "step": 15620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.884762441543407e-05,
+      "loss": 1.4066,
+      "step": 15630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8846862766006064e-05,
+      "loss": 1.3825,
+      "step": 15640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.884610111657806e-05,
+      "loss": 1.3151,
+      "step": 15650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8845339467150066e-05,
+      "loss": 1.2428,
+      "step": 15660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.884457781772206e-05,
+      "loss": 1.2916,
+      "step": 15670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.884381616829406e-05,
+      "loss": 1.3546,
+      "step": 15680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.884305451886606e-05,
+      "loss": 1.2898,
+      "step": 15690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8842292869438056e-05,
+      "loss": 1.3648,
+      "step": 15700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8841531220010053e-05,
+      "loss": 1.2609,
+      "step": 15710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.884076957058206e-05,
+      "loss": 1.2824,
+      "step": 15720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.884000792115405e-05,
+      "loss": 1.4524,
+      "step": 15730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883924627172605e-05,
+      "loss": 1.267,
+      "step": 15740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883848462229805e-05,
+      "loss": 1.3622,
+      "step": 15750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883772297287005e-05,
+      "loss": 1.26,
+      "step": 15760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883696132344205e-05,
+      "loss": 1.2608,
+      "step": 15770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883619967401405e-05,
+      "loss": 1.2496,
+      "step": 15780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883543802458605e-05,
+      "loss": 1.3911,
+      "step": 15790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8834676375158044e-05,
+      "loss": 1.2194,
+      "step": 15800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883391472573005e-05,
+      "loss": 1.3002,
+      "step": 15810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883315307630204e-05,
+      "loss": 1.3094,
+      "step": 15820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883239142687404e-05,
+      "loss": 1.1543,
+      "step": 15830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883162977744604e-05,
+      "loss": 1.3273,
+      "step": 15840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.883086812801804e-05,
+      "loss": 1.2731,
+      "step": 15850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8830106478590036e-05,
+      "loss": 1.2317,
+      "step": 15860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.882934482916204e-05,
+      "loss": 1.3657,
+      "step": 15870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882858317973403e-05,
+      "loss": 1.2196,
+      "step": 15880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8827821530306035e-05,
+      "loss": 1.1452,
+      "step": 15890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882705988087803e-05,
+      "loss": 1.4502,
+      "step": 15900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882629823145003e-05,
+      "loss": 1.2485,
+      "step": 15910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882553658202203e-05,
+      "loss": 1.321,
+      "step": 15920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882477493259403e-05,
+      "loss": 1.4079,
+      "step": 15930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882401328316603e-05,
+      "loss": 1.3802,
+      "step": 15940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8823251633738027e-05,
+      "loss": 1.3484,
+      "step": 15950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8822489984310024e-05,
+      "loss": 1.5103,
+      "step": 15960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.882172833488202e-05,
+      "loss": 1.3559,
+      "step": 15970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8820966685454026e-05,
+      "loss": 1.2201,
+      "step": 15980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8820205036026016e-05,
+      "loss": 1.2207,
+      "step": 15990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881944338659802e-05,
+      "loss": 1.2563,
+      "step": 16000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881868173717002e-05,
+      "loss": 1.3851,
+      "step": 16010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8817920087742016e-05,
+      "loss": 1.3262,
+      "step": 16020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881715843831401e-05,
+      "loss": 1.2929,
+      "step": 16030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881639678888602e-05,
+      "loss": 1.2515,
+      "step": 16040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881563513945801e-05,
+      "loss": 1.3496,
+      "step": 16050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881487349003001e-05,
+      "loss": 1.306,
+      "step": 16060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881411184060201e-05,
+      "loss": 1.3263,
+      "step": 16070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.881335019117401e-05,
+      "loss": 1.2374,
+      "step": 16080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8812588541746005e-05,
+      "loss": 1.3019,
+      "step": 16090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.881182689231801e-05,
+      "loss": 1.2186,
+      "step": 16100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.881106524289e-05,
+      "loss": 1.4081,
+      "step": 16110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8810303593462004e-05,
+      "loss": 1.2133,
+      "step": 16120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8809541944034e-05,
+      "loss": 1.299,
+      "step": 16130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8808780294606e-05,
+      "loss": 1.3202,
+      "step": 16140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8808018645178e-05,
+      "loss": 1.35,
+      "step": 16150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880725699575e-05,
+      "loss": 1.2883,
+      "step": 16160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8806495346322e-05,
+      "loss": 1.5032,
+      "step": 16170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8805733696893995e-05,
+      "loss": 1.273,
+      "step": 16180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8804972047466e-05,
+      "loss": 1.1952,
+      "step": 16190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880421039803799e-05,
+      "loss": 1.2914,
+      "step": 16200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8803448748609995e-05,
+      "loss": 1.3544,
+      "step": 16210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880268709918199e-05,
+      "loss": 1.3244,
+      "step": 16220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880192544975399e-05,
+      "loss": 1.211,
+      "step": 16230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880116380032599e-05,
+      "loss": 1.2914,
+      "step": 16240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.880040215089799e-05,
+      "loss": 1.4709,
+      "step": 16250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.879964050146998e-05,
+      "loss": 1.1832,
+      "step": 16260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8798878852041986e-05,
+      "loss": 1.2657,
+      "step": 16270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8798117202613984e-05,
+      "loss": 1.2407,
+      "step": 16280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.879735555318598e-05,
+      "loss": 1.3274,
+      "step": 16290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.879659390375798e-05,
+      "loss": 1.4302,
+      "step": 16300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.879583225432998e-05,
+      "loss": 1.2953,
+      "step": 16310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8795070604901973e-05,
+      "loss": 1.385,
+      "step": 16320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.879430895547398e-05,
+      "loss": 1.2032,
+      "step": 16330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8793547306045975e-05,
+      "loss": 1.2663,
+      "step": 16340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.879278565661797e-05,
+      "loss": 1.4407,
+      "step": 16350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.879202400718998e-05,
+      "loss": 1.458,
+      "step": 16360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8791262357761974e-05,
+      "loss": 1.2033,
+      "step": 16370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.879050070833397e-05,
+      "loss": 1.301,
+      "step": 16380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878973905890597e-05,
+      "loss": 1.2842,
+      "step": 16390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8788977409477974e-05,
+      "loss": 1.3248,
+      "step": 16400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8788215760049964e-05,
+      "loss": 1.2152,
+      "step": 16410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878745411062197e-05,
+      "loss": 1.3181,
+      "step": 16420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878669246119396e-05,
+      "loss": 1.3743,
+      "step": 16430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8785930811765963e-05,
+      "loss": 1.1879,
+      "step": 16440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878516916233796e-05,
+      "loss": 1.265,
+      "step": 16450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878440751290996e-05,
+      "loss": 1.3624,
+      "step": 16460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8783645863481956e-05,
+      "loss": 1.3496,
+      "step": 16470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878288421405396e-05,
+      "loss": 1.3833,
+      "step": 16480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878212256462595e-05,
+      "loss": 1.3073,
+      "step": 16490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8781360915197955e-05,
+      "loss": 1.141,
+      "step": 16500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.878059926576995e-05,
+      "loss": 1.3386,
+      "step": 16510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.877983761634195e-05,
+      "loss": 1.4003,
+      "step": 16520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8779075966913954e-05,
+      "loss": 1.3143,
+      "step": 16530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877831431748595e-05,
+      "loss": 1.4422,
+      "step": 16540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877755266805795e-05,
+      "loss": 1.1821,
+      "step": 16550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8776791018629947e-05,
+      "loss": 1.1904,
+      "step": 16560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877602936920195e-05,
+      "loss": 1.1895,
+      "step": 16570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877526771977394e-05,
+      "loss": 1.2275,
+      "step": 16580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8774506070345946e-05,
+      "loss": 1.3871,
+      "step": 16590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877374442091794e-05,
+      "loss": 1.489,
+      "step": 16600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877298277148994e-05,
+      "loss": 1.1939,
+      "step": 16610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877222112206194e-05,
+      "loss": 1.2439,
+      "step": 16620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877145947263394e-05,
+      "loss": 1.2424,
+      "step": 16630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877069782320593e-05,
+      "loss": 1.3432,
+      "step": 16640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876993617377794e-05,
+      "loss": 1.3057,
+      "step": 16650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8769174524349935e-05,
+      "loss": 1.3642,
+      "step": 16660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876841287492193e-05,
+      "loss": 1.3029,
+      "step": 16670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876765122549393e-05,
+      "loss": 1.4206,
+      "step": 16680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8766889576065934e-05,
+      "loss": 1.3407,
+      "step": 16690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8766127926637925e-05,
+      "loss": 1.3989,
+      "step": 16700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876536627720993e-05,
+      "loss": 1.359,
+      "step": 16710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8764604627781926e-05,
+      "loss": 1.5363,
+      "step": 16720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8763842978353924e-05,
+      "loss": 1.4084,
+      "step": 16730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876308132892593e-05,
+      "loss": 1.2201,
+      "step": 16740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8762319679497926e-05,
+      "loss": 1.4777,
+      "step": 16750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.876155803006992e-05,
+      "loss": 1.3036,
+      "step": 16760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.876079638064192e-05,
+      "loss": 1.1362,
+      "step": 16770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8760034731213925e-05,
+      "loss": 1.2673,
+      "step": 16780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8759273081785915e-05,
+      "loss": 1.3441,
+      "step": 16790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.875851143235792e-05,
+      "loss": 1.2253,
+      "step": 16800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.875774978292992e-05,
+      "loss": 1.3199,
+      "step": 16810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8756988133501915e-05,
+      "loss": 1.2476,
+      "step": 16820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.875622648407391e-05,
+      "loss": 1.2848,
+      "step": 16830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8755464834645916e-05,
+      "loss": 1.2047,
+      "step": 16840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.875470318521791e-05,
+      "loss": 1.299,
+      "step": 16850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.875394153578991e-05,
+      "loss": 1.1488,
+      "step": 16860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87531798863619e-05,
+      "loss": 1.3498,
+      "step": 16870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8752418236933906e-05,
+      "loss": 1.2336,
+      "step": 16880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8751656587505904e-05,
+      "loss": 1.3848,
+      "step": 16890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87508949380779e-05,
+      "loss": 1.2154,
+      "step": 16900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87501332886499e-05,
+      "loss": 1.3458,
+      "step": 16910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87493716392219e-05,
+      "loss": 1.2699,
+      "step": 16920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87486099897939e-05,
+      "loss": 1.3101,
+      "step": 16930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87478483403659e-05,
+      "loss": 1.2349,
+      "step": 16940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87470866909379e-05,
+      "loss": 1.3245,
+      "step": 16950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.874632504150989e-05,
+      "loss": 1.2697,
+      "step": 16960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87455633920819e-05,
+      "loss": 1.2701,
+      "step": 16970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8744801742653894e-05,
+      "loss": 1.2495,
+      "step": 16980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.874404009322589e-05,
+      "loss": 1.4603,
+      "step": 16990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.874327844379789e-05,
+      "loss": 1.5377,
+      "step": 17000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8742516794369894e-05,
+      "loss": 1.255,
+      "step": 17010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8741755144941884e-05,
+      "loss": 1.3455,
+      "step": 17020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.874099349551389e-05,
+      "loss": 1.5043,
+      "step": 17030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8740231846085886e-05,
+      "loss": 1.3453,
+      "step": 17040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8739470196657883e-05,
+      "loss": 1.4523,
+      "step": 17050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873870854722988e-05,
+      "loss": 1.2793,
+      "step": 17060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8737946897801885e-05,
+      "loss": 1.3737,
+      "step": 17070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8737185248373876e-05,
+      "loss": 1.4294,
+      "step": 17080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873642359894588e-05,
+      "loss": 1.2856,
+      "step": 17090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873566194951788e-05,
+      "loss": 1.213,
+      "step": 17100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8734900300089875e-05,
+      "loss": 1.2516,
+      "step": 17110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873413865066187e-05,
+      "loss": 1.3498,
+      "step": 17120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873337700123388e-05,
+      "loss": 1.3161,
+      "step": 17130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8732615351805874e-05,
+      "loss": 1.2401,
+      "step": 17140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.873185370237787e-05,
+      "loss": 1.2127,
+      "step": 17150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8731092052949876e-05,
+      "loss": 1.243,
+      "step": 17160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8730330403521867e-05,
+      "loss": 1.2494,
+      "step": 17170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.872956875409387e-05,
+      "loss": 1.2642,
+      "step": 17180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.872880710466587e-05,
+      "loss": 1.3268,
+      "step": 17190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8728045455237866e-05,
+      "loss": 1.3388,
+      "step": 17200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872728380580986e-05,
+      "loss": 1.4205,
+      "step": 17210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872652215638187e-05,
+      "loss": 1.2039,
+      "step": 17220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872576050695386e-05,
+      "loss": 1.2138,
+      "step": 17230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872499885752586e-05,
+      "loss": 1.3251,
+      "step": 17240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872423720809786e-05,
+      "loss": 1.3405,
+      "step": 17250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872347555866986e-05,
+      "loss": 1.211,
+      "step": 17260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8722713909241855e-05,
+      "loss": 1.1951,
+      "step": 17270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872195225981386e-05,
+      "loss": 1.3661,
+      "step": 17280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.872119061038585e-05,
+      "loss": 1.2919,
+      "step": 17290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8720428960957854e-05,
+      "loss": 1.1753,
+      "step": 17300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871966731152985e-05,
+      "loss": 1.227,
+      "step": 17310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871890566210185e-05,
+      "loss": 1.2825,
+      "step": 17320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871814401267385e-05,
+      "loss": 1.1973,
+      "step": 17330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8717382363245844e-05,
+      "loss": 1.4051,
+      "step": 17340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871662071381785e-05,
+      "loss": 1.2347,
+      "step": 17350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8715859064389846e-05,
+      "loss": 1.2212,
+      "step": 17360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871509741496184e-05,
+      "loss": 1.337,
+      "step": 17370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871433576553384e-05,
+      "loss": 1.2877,
+      "step": 17380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8713574116105845e-05,
+      "loss": 1.3652,
+      "step": 17390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8712812466677835e-05,
+      "loss": 1.3491,
+      "step": 17400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.871205081724984e-05,
+      "loss": 1.1322,
+      "step": 17410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.871128916782184e-05,
+      "loss": 1.312,
+      "step": 17420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8710527518393835e-05,
+      "loss": 1.1455,
+      "step": 17430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870976586896583e-05,
+      "loss": 1.362,
+      "step": 17440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8709004219537836e-05,
+      "loss": 1.2271,
+      "step": 17450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870824257010983e-05,
+      "loss": 1.3924,
+      "step": 17460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870748092068183e-05,
+      "loss": 1.3803,
+      "step": 17470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870671927125383e-05,
+      "loss": 1.2692,
+      "step": 17480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8705957621825826e-05,
+      "loss": 1.3235,
+      "step": 17490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8705195972397824e-05,
+      "loss": 1.2892,
+      "step": 17500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870443432296983e-05,
+      "loss": 1.2679,
+      "step": 17510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8703672673541825e-05,
+      "loss": 1.3869,
+      "step": 17520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870291102411382e-05,
+      "loss": 1.2889,
+      "step": 17530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870214937468583e-05,
+      "loss": 1.2946,
+      "step": 17540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870138772525782e-05,
+      "loss": 1.1785,
+      "step": 17550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870062607582982e-05,
+      "loss": 1.4284,
+      "step": 17560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.869986442640182e-05,
+      "loss": 1.1118,
+      "step": 17570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.869910277697382e-05,
+      "loss": 1.3923,
+      "step": 17580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8698341127545814e-05,
+      "loss": 1.1514,
+      "step": 17590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.869757947811782e-05,
+      "loss": 1.2181,
+      "step": 17600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.869681782868981e-05,
+      "loss": 1.3533,
+      "step": 17610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8696056179261814e-05,
+      "loss": 1.4259,
+      "step": 17620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.869529452983381e-05,
+      "loss": 1.1769,
+      "step": 17630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.869453288040581e-05,
+      "loss": 1.1697,
+      "step": 17640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8693771230977806e-05,
+      "loss": 1.2466,
+      "step": 17650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.869300958154981e-05,
+      "loss": 1.3961,
+      "step": 17660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86922479321218e-05,
+      "loss": 1.1935,
+      "step": 17670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8691486282693805e-05,
+      "loss": 1.3094,
+      "step": 17680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86907246332658e-05,
+      "loss": 1.368,
+      "step": 17690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86899629838378e-05,
+      "loss": 1.2846,
+      "step": 17700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86892013344098e-05,
+      "loss": 1.2304,
+      "step": 17710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86884396849818e-05,
+      "loss": 1.185,
+      "step": 17720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86876780355538e-05,
+      "loss": 1.4301,
+      "step": 17730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86869163861258e-05,
+      "loss": 1.2269,
+      "step": 17740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86861547366978e-05,
+      "loss": 1.3311,
+      "step": 17750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.868539308726979e-05,
+      "loss": 1.2239,
+      "step": 17760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8684631437841796e-05,
+      "loss": 1.2854,
+      "step": 17770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.868386978841379e-05,
+      "loss": 1.2501,
+      "step": 17780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.868310813898579e-05,
+      "loss": 1.1171,
+      "step": 17790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.868234648955779e-05,
+      "loss": 1.2943,
+      "step": 17800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8681584840129786e-05,
+      "loss": 1.364,
+      "step": 17810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.868082319070178e-05,
+      "loss": 1.244,
+      "step": 17820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.868006154127379e-05,
+      "loss": 1.326,
+      "step": 17830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.867929989184578e-05,
+      "loss": 1.3356,
+      "step": 17840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867853824241778e-05,
+      "loss": 1.1696,
+      "step": 17850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867777659298978e-05,
+      "loss": 1.2671,
+      "step": 17860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867701494356178e-05,
+      "loss": 1.4618,
+      "step": 17870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8676253294133775e-05,
+      "loss": 1.3042,
+      "step": 17880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867549164470578e-05,
+      "loss": 1.1933,
+      "step": 17890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8674729995277777e-05,
+      "loss": 1.3285,
+      "step": 17900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8673968345849774e-05,
+      "loss": 1.2823,
+      "step": 17910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867320669642177e-05,
+      "loss": 1.4067,
+      "step": 17920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867244504699377e-05,
+      "loss": 1.2996,
+      "step": 17930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867168339756577e-05,
+      "loss": 1.4401,
+      "step": 17940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867092174813777e-05,
+      "loss": 1.3589,
+      "step": 17950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.867016009870977e-05,
+      "loss": 1.0869,
+      "step": 17960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8669398449281766e-05,
+      "loss": 1.1839,
+      "step": 17970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866863679985377e-05,
+      "loss": 1.344,
+      "step": 17980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866787515042576e-05,
+      "loss": 1.4928,
+      "step": 17990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8667113500997765e-05,
+      "loss": 1.3393,
+      "step": 18000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866635185156976e-05,
+      "loss": 1.2806,
+      "step": 18010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866559020214176e-05,
+      "loss": 1.4632,
+      "step": 18020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866482855271376e-05,
+      "loss": 1.1801,
+      "step": 18030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866406690328576e-05,
+      "loss": 1.3823,
+      "step": 18040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866330525385775e-05,
+      "loss": 1.3501,
+      "step": 18050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8662543604429756e-05,
+      "loss": 1.346,
+      "step": 18060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8661781955001754e-05,
+      "loss": 1.368,
+      "step": 18070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.866102030557375e-05,
+      "loss": 1.4265,
+      "step": 18080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.866025865614575e-05,
+      "loss": 1.1587,
+      "step": 18090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865949700671775e-05,
+      "loss": 1.2246,
+      "step": 18100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865873535728975e-05,
+      "loss": 1.5215,
+      "step": 18110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865797370786175e-05,
+      "loss": 1.3174,
+      "step": 18120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865721205843375e-05,
+      "loss": 1.3275,
+      "step": 18130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865645040900574e-05,
+      "loss": 1.4369,
+      "step": 18140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865568875957775e-05,
+      "loss": 1.1185,
+      "step": 18150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8654927110149745e-05,
+      "loss": 1.2205,
+      "step": 18160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865416546072174e-05,
+      "loss": 1.3108,
+      "step": 18170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865340381129374e-05,
+      "loss": 1.2196,
+      "step": 18180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8652642161865744e-05,
+      "loss": 1.3494,
+      "step": 18190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8651880512437734e-05,
+      "loss": 1.3597,
+      "step": 18200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.865111886300974e-05,
+      "loss": 1.3421,
+      "step": 18210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8650357213581736e-05,
+      "loss": 1.2513,
+      "step": 18220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8649595564153734e-05,
+      "loss": 1.4412,
+      "step": 18230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.864883391472573e-05,
+      "loss": 1.3759,
+      "step": 18240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.864807226529773e-05,
+      "loss": 1.3984,
+      "step": 18250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8647310615869726e-05,
+      "loss": 1.1095,
+      "step": 18260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.864654896644173e-05,
+      "loss": 1.321,
+      "step": 18270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.864578731701373e-05,
+      "loss": 1.1788,
+      "step": 18280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8645025667585725e-05,
+      "loss": 1.2716,
+      "step": 18290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.864426401815772e-05,
+      "loss": 1.373,
+      "step": 18300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.864350236872972e-05,
+      "loss": 1.295,
+      "step": 18310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8642740719301724e-05,
+      "loss": 1.3211,
+      "step": 18320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.864197906987372e-05,
+      "loss": 1.3217,
+      "step": 18330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.864121742044572e-05,
+      "loss": 1.3061,
+      "step": 18340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.864045577101772e-05,
+      "loss": 1.3692,
+      "step": 18350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863969412158972e-05,
+      "loss": 1.1817,
+      "step": 18360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863893247216171e-05,
+      "loss": 1.3354,
+      "step": 18370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8638170822733716e-05,
+      "loss": 1.2987,
+      "step": 18380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863740917330571e-05,
+      "loss": 1.1509,
+      "step": 18390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863664752387771e-05,
+      "loss": 1.2111,
+      "step": 18400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863588587444971e-05,
+      "loss": 1.3202,
+      "step": 18410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863512422502171e-05,
+      "loss": 1.1793,
+      "step": 18420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.86343625755937e-05,
+      "loss": 1.2699,
+      "step": 18430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.863360092616571e-05,
+      "loss": 1.2003,
+      "step": 18440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8632839276737705e-05,
+      "loss": 1.31,
+      "step": 18450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.86320776273097e-05,
+      "loss": 1.374,
+      "step": 18460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.86313159778817e-05,
+      "loss": 1.3333,
+      "step": 18470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8630554328453704e-05,
+      "loss": 1.3983,
+      "step": 18480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.86297926790257e-05,
+      "loss": 1.2187,
+      "step": 18490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.86290310295977e-05,
+      "loss": 1.2351,
+      "step": 18500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8628269380169697e-05,
+      "loss": 1.17,
+      "step": 18510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8627507730741694e-05,
+      "loss": 1.2563,
+      "step": 18520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.86267460813137e-05,
+      "loss": 1.3336,
+      "step": 18530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8625984431885696e-05,
+      "loss": 1.2312,
+      "step": 18540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.862522278245769e-05,
+      "loss": 1.2209,
+      "step": 18550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.862446113302969e-05,
+      "loss": 1.2724,
+      "step": 18560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8623699483601695e-05,
+      "loss": 1.4351,
+      "step": 18570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8622937834173686e-05,
+      "loss": 1.3779,
+      "step": 18580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.862217618474569e-05,
+      "loss": 1.3168,
+      "step": 18590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.862141453531769e-05,
+      "loss": 1.2049,
+      "step": 18600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8620652885889685e-05,
+      "loss": 1.2944,
+      "step": 18610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861989123646168e-05,
+      "loss": 1.2876,
+      "step": 18620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8619129587033686e-05,
+      "loss": 1.3433,
+      "step": 18630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861836793760568e-05,
+      "loss": 1.1461,
+      "step": 18640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861760628817768e-05,
+      "loss": 1.3349,
+      "step": 18650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861684463874968e-05,
+      "loss": 1.3615,
+      "step": 18660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8616082989321676e-05,
+      "loss": 1.2972,
+      "step": 18670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8615321339893674e-05,
+      "loss": 1.2885,
+      "step": 18680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861455969046568e-05,
+      "loss": 1.1745,
+      "step": 18690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8613798041037675e-05,
+      "loss": 1.3293,
+      "step": 18700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861303639160967e-05,
+      "loss": 1.3454,
+      "step": 18710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.861227474218167e-05,
+      "loss": 1.2738,
+      "step": 18720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.861151309275367e-05,
+      "loss": 1.2082,
+      "step": 18730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.861075144332567e-05,
+      "loss": 1.2617,
+      "step": 18740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860998979389766e-05,
+      "loss": 1.3662,
+      "step": 18750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860922814446967e-05,
+      "loss": 1.2606,
+      "step": 18760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8608466495041665e-05,
+      "loss": 1.1312,
+      "step": 18770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860770484561366e-05,
+      "loss": 1.2892,
+      "step": 18780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860694319618566e-05,
+      "loss": 1.3055,
+      "step": 18790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8606181546757664e-05,
+      "loss": 1.286,
+      "step": 18800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8605419897329654e-05,
+      "loss": 1.322,
+      "step": 18810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860465824790166e-05,
+      "loss": 1.2606,
+      "step": 18820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8603896598473656e-05,
+      "loss": 1.1813,
+      "step": 18830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8603134949045654e-05,
+      "loss": 1.3994,
+      "step": 18840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860237329961765e-05,
+      "loss": 1.1781,
+      "step": 18850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8601611650189655e-05,
+      "loss": 1.2706,
+      "step": 18860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860085000076165e-05,
+      "loss": 1.2517,
+      "step": 18870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860008835133365e-05,
+      "loss": 1.1548,
+      "step": 18880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.859932670190565e-05,
+      "loss": 1.1711,
+      "step": 18890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8598565052477645e-05,
+      "loss": 1.2491,
+      "step": 18900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.859780340304965e-05,
+      "loss": 1.2919,
+      "step": 18910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.859704175362165e-05,
+      "loss": 1.2922,
+      "step": 18920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8596280104193644e-05,
+      "loss": 1.3753,
+      "step": 18930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.859551845476564e-05,
+      "loss": 1.2588,
+      "step": 18940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8594756805337646e-05,
+      "loss": 1.2791,
+      "step": 18950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.859399515590964e-05,
+      "loss": 1.286,
+      "step": 18960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.859323350648164e-05,
+      "loss": 1.2287,
+      "step": 18970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.859247185705364e-05,
+      "loss": 1.2914,
+      "step": 18980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8591710207625636e-05,
+      "loss": 1.1893,
+      "step": 18990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.859094855819763e-05,
+      "loss": 1.138,
+      "step": 19000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.859018690876964e-05,
+      "loss": 1.3644,
+      "step": 19010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858942525934163e-05,
+      "loss": 1.3435,
+      "step": 19020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858866360991363e-05,
+      "loss": 1.2937,
+      "step": 19030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858790196048563e-05,
+      "loss": 1.449,
+      "step": 19040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858714031105763e-05,
+      "loss": 1.3438,
+      "step": 19050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8586378661629625e-05,
+      "loss": 1.1988,
+      "step": 19060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858561701220163e-05,
+      "loss": 1.2602,
+      "step": 19070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858485536277363e-05,
+      "loss": 1.3143,
+      "step": 19080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8584093713345624e-05,
+      "loss": 1.1819,
+      "step": 19090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858333206391762e-05,
+      "loss": 1.1687,
+      "step": 19100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858257041448962e-05,
+      "loss": 1.3369,
+      "step": 19110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858180876506162e-05,
+      "loss": 1.2457,
+      "step": 19120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858104711563362e-05,
+      "loss": 1.3189,
+      "step": 19130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858028546620562e-05,
+      "loss": 1.1733,
+      "step": 19140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8579523816777616e-05,
+      "loss": 1.3352,
+      "step": 19150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.857876216734962e-05,
+      "loss": 1.1931,
+      "step": 19160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.857800051792161e-05,
+      "loss": 1.3779,
+      "step": 19170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8577238868493615e-05,
+      "loss": 1.2343,
+      "step": 19180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8576477219065606e-05,
+      "loss": 1.1419,
+      "step": 19190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.857571556963761e-05,
+      "loss": 1.2738,
+      "step": 19200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.857495392020961e-05,
+      "loss": 1.3001,
+      "step": 19210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8574192270781605e-05,
+      "loss": 1.3236,
+      "step": 19220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85734306213536e-05,
+      "loss": 1.393,
+      "step": 19230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8572668971925606e-05,
+      "loss": 1.1973,
+      "step": 19240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8571907322497604e-05,
+      "loss": 1.2519,
+      "step": 19250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85711456730696e-05,
+      "loss": 1.3252,
+      "step": 19260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85703840236416e-05,
+      "loss": 1.3319,
+      "step": 19270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8569622374213596e-05,
+      "loss": 1.2305,
+      "step": 19280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85688607247856e-05,
+      "loss": 1.219,
+      "step": 19290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85680990753576e-05,
+      "loss": 1.3394,
+      "step": 19300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8567337425929595e-05,
+      "loss": 1.2785,
+      "step": 19310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.856657577650159e-05,
+      "loss": 1.4561,
+      "step": 19320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85658141270736e-05,
+      "loss": 1.305,
+      "step": 19330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.856505247764559e-05,
+      "loss": 1.1348,
+      "step": 19340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.856429082821759e-05,
+      "loss": 1.2873,
+      "step": 19350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.856352917878959e-05,
+      "loss": 1.1991,
+      "step": 19360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.856276752936159e-05,
+      "loss": 1.2911,
+      "step": 19370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8562005879933585e-05,
+      "loss": 1.2752,
+      "step": 19380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.856124423050559e-05,
+      "loss": 1.3918,
+      "step": 19390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.856048258107758e-05,
+      "loss": 1.2754,
+      "step": 19400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8559720931649584e-05,
+      "loss": 1.2223,
+      "step": 19410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855895928222158e-05,
+      "loss": 1.3121,
+      "step": 19420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855819763279358e-05,
+      "loss": 1.3199,
+      "step": 19430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8557435983365576e-05,
+      "loss": 1.2025,
+      "step": 19440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855667433393758e-05,
+      "loss": 1.2969,
+      "step": 19450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855591268450958e-05,
+      "loss": 1.2334,
+      "step": 19460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8555151035081575e-05,
+      "loss": 1.235,
+      "step": 19470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855438938565357e-05,
+      "loss": 1.1175,
+      "step": 19480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855362773622557e-05,
+      "loss": 1.1419,
+      "step": 19490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8552866086797574e-05,
+      "loss": 1.3926,
+      "step": 19500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855210443736957e-05,
+      "loss": 1.3191,
+      "step": 19510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855134278794157e-05,
+      "loss": 1.2501,
+      "step": 19520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.855058113851357e-05,
+      "loss": 1.0919,
+      "step": 19530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.854981948908557e-05,
+      "loss": 1.0906,
+      "step": 19540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.854905783965756e-05,
+      "loss": 1.2107,
+      "step": 19550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8548296190229566e-05,
+      "loss": 1.3479,
+      "step": 19560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8547534540801564e-05,
+      "loss": 1.1323,
+      "step": 19570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.854677289137356e-05,
+      "loss": 1.3267,
+      "step": 19580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.854601124194556e-05,
+      "loss": 1.3161,
+      "step": 19590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.854524959251756e-05,
+      "loss": 1.2798,
+      "step": 19600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854448794308955e-05,
+      "loss": 1.2525,
+      "step": 19610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854372629366156e-05,
+      "loss": 1.1452,
+      "step": 19620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8542964644233555e-05,
+      "loss": 1.2972,
+      "step": 19630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854220299480555e-05,
+      "loss": 1.4298,
+      "step": 19640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854144134537755e-05,
+      "loss": 1.2573,
+      "step": 19650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854067969594955e-05,
+      "loss": 1.3387,
+      "step": 19660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853991804652155e-05,
+      "loss": 1.1492,
+      "step": 19670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853915639709355e-05,
+      "loss": 1.1941,
+      "step": 19680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853839474766555e-05,
+      "loss": 1.406,
+      "step": 19690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8537633098237544e-05,
+      "loss": 1.3235,
+      "step": 19700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853687144880955e-05,
+      "loss": 1.2957,
+      "step": 19710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853610979938154e-05,
+      "loss": 1.2488,
+      "step": 19720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853534814995354e-05,
+      "loss": 1.258,
+      "step": 19730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853458650052554e-05,
+      "loss": 1.2884,
+      "step": 19740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853382485109754e-05,
+      "loss": 1.3429,
+      "step": 19750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8533063201669536e-05,
+      "loss": 1.1756,
+      "step": 19760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853230155224154e-05,
+      "loss": 1.3358,
+      "step": 19770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853153990281353e-05,
+      "loss": 1.2126,
+      "step": 19780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8530778253385535e-05,
+      "loss": 1.0889,
+      "step": 19790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853001660395753e-05,
+      "loss": 1.2895,
+      "step": 19800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852925495452953e-05,
+      "loss": 1.2654,
+      "step": 19810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852849330510153e-05,
+      "loss": 1.2133,
+      "step": 19820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852773165567353e-05,
+      "loss": 1.2021,
+      "step": 19830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852697000624553e-05,
+      "loss": 1.35,
+      "step": 19840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8526208356817526e-05,
+      "loss": 1.3609,
+      "step": 19850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8525446707389524e-05,
+      "loss": 1.3516,
+      "step": 19860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852468505796152e-05,
+      "loss": 1.3684,
+      "step": 19870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8523923408533526e-05,
+      "loss": 1.2515,
+      "step": 19880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852316175910552e-05,
+      "loss": 1.3244,
+      "step": 19890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852240010967752e-05,
+      "loss": 1.2037,
+      "step": 19900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852163846024952e-05,
+      "loss": 1.3044,
+      "step": 19910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852087681082152e-05,
+      "loss": 1.3398,
+      "step": 19920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.852011516139351e-05,
+      "loss": 1.3301,
+      "step": 19930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.851935351196552e-05,
+      "loss": 1.3079,
+      "step": 19940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8518591862537515e-05,
+      "loss": 1.2446,
+      "step": 19950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.851783021310951e-05,
+      "loss": 1.3781,
+      "step": 19960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.851706856368151e-05,
+      "loss": 1.3016,
+      "step": 19970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8516306914253514e-05,
+      "loss": 1.2244,
+      "step": 19980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8515545264825505e-05,
+      "loss": 1.2637,
+      "step": 19990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.851478361539751e-05,
+      "loss": 1.3225,
+      "step": 20000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8514021965969506e-05,
+      "loss": 1.295,
+      "step": 20010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8513260316541504e-05,
+      "loss": 1.2491,
+      "step": 20020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.85124986671135e-05,
+      "loss": 1.2401,
+      "step": 20030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8511737017685505e-05,
+      "loss": 1.4517,
+      "step": 20040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.85109753682575e-05,
+      "loss": 1.2419,
+      "step": 20050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.85102137188295e-05,
+      "loss": 1.1633,
+      "step": 20060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.85094520694015e-05,
+      "loss": 1.1734,
+      "step": 20070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8508690419973495e-05,
+      "loss": 1.335,
+      "step": 20080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.85079287705455e-05,
+      "loss": 1.2064,
+      "step": 20090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850716712111749e-05,
+      "loss": 1.2283,
+      "step": 20100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8506405471689494e-05,
+      "loss": 1.2793,
+      "step": 20110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850564382226149e-05,
+      "loss": 1.233,
+      "step": 20120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850488217283349e-05,
+      "loss": 1.1635,
+      "step": 20130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850412052340549e-05,
+      "loss": 1.2976,
+      "step": 20140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850335887397749e-05,
+      "loss": 1.163,
+      "step": 20150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850259722454948e-05,
+      "loss": 1.3045,
+      "step": 20160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8501835575121486e-05,
+      "loss": 1.2991,
+      "step": 20170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8501073925693484e-05,
+      "loss": 1.3121,
+      "step": 20180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.850031227626548e-05,
+      "loss": 1.3149,
+      "step": 20190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.849955062683748e-05,
+      "loss": 1.3927,
+      "step": 20200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.849878897740948e-05,
+      "loss": 1.3218,
+      "step": 20210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8498027327981473e-05,
+      "loss": 1.261,
+      "step": 20220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.849726567855348e-05,
+      "loss": 1.1999,
+      "step": 20230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8496504029125475e-05,
+      "loss": 1.3708,
+      "step": 20240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.849574237969747e-05,
+      "loss": 1.2293,
+      "step": 20250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849498073026948e-05,
+      "loss": 1.2703,
+      "step": 20260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8494219080841474e-05,
+      "loss": 1.2389,
+      "step": 20270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849345743141347e-05,
+      "loss": 1.0818,
+      "step": 20280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849269578198547e-05,
+      "loss": 1.1331,
+      "step": 20290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8491934132557473e-05,
+      "loss": 1.3625,
+      "step": 20300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8491172483129464e-05,
+      "loss": 1.2325,
+      "step": 20310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849041083370147e-05,
+      "loss": 1.2022,
+      "step": 20320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8489649184273466e-05,
+      "loss": 1.2389,
+      "step": 20330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848888753484546e-05,
+      "loss": 1.531,
+      "step": 20340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848812588541746e-05,
+      "loss": 1.4523,
+      "step": 20350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8487364235989465e-05,
+      "loss": 1.3423,
+      "step": 20360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8486602586561456e-05,
+      "loss": 1.3075,
+      "step": 20370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848584093713346e-05,
+      "loss": 1.2476,
+      "step": 20380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848507928770546e-05,
+      "loss": 1.2993,
+      "step": 20390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8484317638277455e-05,
+      "loss": 1.3288,
+      "step": 20400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848355598884945e-05,
+      "loss": 1.2937,
+      "step": 20410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8482794339421457e-05,
+      "loss": 1.2955,
+      "step": 20420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8482032689993454e-05,
+      "loss": 1.4273,
+      "step": 20430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848127104056545e-05,
+      "loss": 1.1692,
+      "step": 20440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848050939113745e-05,
+      "loss": 1.2904,
+      "step": 20450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8479747741709446e-05,
+      "loss": 1.298,
+      "step": 20460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.847898609228145e-05,
+      "loss": 1.2009,
+      "step": 20470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847822444285345e-05,
+      "loss": 1.2316,
+      "step": 20480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8477462793425446e-05,
+      "loss": 1.2666,
+      "step": 20490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847670114399744e-05,
+      "loss": 1.2662,
+      "step": 20500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847593949456945e-05,
+      "loss": 1.2729,
+      "step": 20510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847517784514144e-05,
+      "loss": 1.2403,
+      "step": 20520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847441619571344e-05,
+      "loss": 1.2107,
+      "step": 20530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847365454628544e-05,
+      "loss": 1.1908,
+      "step": 20540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847289289685744e-05,
+      "loss": 1.2147,
+      "step": 20550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8472131247429435e-05,
+      "loss": 1.1768,
+      "step": 20560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847136959800143e-05,
+      "loss": 1.3349,
+      "step": 20570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847060794857343e-05,
+      "loss": 1.3205,
+      "step": 20580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8469846299145434e-05,
+      "loss": 1.1692,
+      "step": 20590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8469084649717425e-05,
+      "loss": 1.1457,
+      "step": 20600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.846832300028943e-05,
+      "loss": 1.2965,
+      "step": 20610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8467561350861426e-05,
+      "loss": 1.3647,
+      "step": 20620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8466799701433424e-05,
+      "loss": 1.4168,
+      "step": 20630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.846603805200543e-05,
+      "loss": 1.2196,
+      "step": 20640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8465276402577425e-05,
+      "loss": 1.2489,
+      "step": 20650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.846451475314942e-05,
+      "loss": 1.2535,
+      "step": 20660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.846375310372142e-05,
+      "loss": 1.2438,
+      "step": 20670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8462991454293425e-05,
+      "loss": 1.1755,
+      "step": 20680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8462229804865415e-05,
+      "loss": 1.1992,
+      "step": 20690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.846146815543742e-05,
+      "loss": 1.1605,
+      "step": 20700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.846070650600942e-05,
+      "loss": 1.2476,
+      "step": 20710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8459944856581414e-05,
+      "loss": 1.3172,
+      "step": 20720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845918320715341e-05,
+      "loss": 1.1539,
+      "step": 20730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8458421557725416e-05,
+      "loss": 1.2799,
+      "step": 20740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845765990829741e-05,
+      "loss": 1.1081,
+      "step": 20750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845689825886941e-05,
+      "loss": 1.195,
+      "step": 20760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845613660944141e-05,
+      "loss": 1.1466,
+      "step": 20770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8455374960013406e-05,
+      "loss": 1.1294,
+      "step": 20780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8454613310585404e-05,
+      "loss": 1.2956,
+      "step": 20790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.845385166115741e-05,
+      "loss": 1.2216,
+      "step": 20800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84530900117294e-05,
+      "loss": 1.197,
+      "step": 20810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84523283623014e-05,
+      "loss": 1.1952,
+      "step": 20820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84515667128734e-05,
+      "loss": 1.1944,
+      "step": 20830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84508050634454e-05,
+      "loss": 1.3143,
+      "step": 20840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84500434140174e-05,
+      "loss": 1.4067,
+      "step": 20850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84492817645894e-05,
+      "loss": 1.3022,
+      "step": 20860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84485201151614e-05,
+      "loss": 1.168,
+      "step": 20870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8447758465733394e-05,
+      "loss": 1.318,
+      "step": 20880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84469968163054e-05,
+      "loss": 1.1023,
+      "step": 20890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.844623516687739e-05,
+      "loss": 1.2486,
+      "step": 20900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8445473517449393e-05,
+      "loss": 1.2297,
+      "step": 20910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.844471186802139e-05,
+      "loss": 1.4166,
+      "step": 20920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.844395021859339e-05,
+      "loss": 1.2636,
+      "step": 20930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8443188569165386e-05,
+      "loss": 1.344,
+      "step": 20940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.844242691973739e-05,
+      "loss": 1.2096,
+      "step": 20950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.844166527030938e-05,
+      "loss": 1.3041,
+      "step": 20960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8440903620881385e-05,
+      "loss": 1.1662,
+      "step": 20970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.844014197145338e-05,
+      "loss": 1.1248,
+      "step": 20980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843938032202538e-05,
+      "loss": 1.396,
+      "step": 20990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843861867259738e-05,
+      "loss": 1.2537,
+      "step": 21000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843785702316938e-05,
+      "loss": 1.0915,
+      "step": 21010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843709537374137e-05,
+      "loss": 1.2531,
+      "step": 21020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843633372431338e-05,
+      "loss": 1.2445,
+      "step": 21030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8435572074885374e-05,
+      "loss": 1.19,
+      "step": 21040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843481042545737e-05,
+      "loss": 1.2118,
+      "step": 21050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8434048776029376e-05,
+      "loss": 1.2392,
+      "step": 21060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8433287126601366e-05,
+      "loss": 1.2878,
+      "step": 21070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843252547717337e-05,
+      "loss": 1.2577,
+      "step": 21080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843176382774537e-05,
+      "loss": 1.3128,
+      "step": 21090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8431002178317366e-05,
+      "loss": 1.2162,
+      "step": 21100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843024052888936e-05,
+      "loss": 1.342,
+      "step": 21110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.842947887946137e-05,
+      "loss": 1.1577,
+      "step": 21120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.842871723003336e-05,
+      "loss": 1.2269,
+      "step": 21130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842795558060536e-05,
+      "loss": 1.4219,
+      "step": 21140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842719393117736e-05,
+      "loss": 1.194,
+      "step": 21150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842643228174936e-05,
+      "loss": 1.3335,
+      "step": 21160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8425670632321355e-05,
+      "loss": 1.2662,
+      "step": 21170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842490898289336e-05,
+      "loss": 1.3127,
+      "step": 21180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842414733346535e-05,
+      "loss": 1.1829,
+      "step": 21190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8423385684037354e-05,
+      "loss": 1.1422,
+      "step": 21200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842262403460935e-05,
+      "loss": 1.2492,
+      "step": 21210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842186238518135e-05,
+      "loss": 1.3845,
+      "step": 21220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842110073575335e-05,
+      "loss": 1.2245,
+      "step": 21230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842033908632535e-05,
+      "loss": 1.1926,
+      "step": 21240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841957743689735e-05,
+      "loss": 1.2479,
+      "step": 21250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8418815787469345e-05,
+      "loss": 1.2927,
+      "step": 21260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841805413804135e-05,
+      "loss": 1.2656,
+      "step": 21270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841729248861334e-05,
+      "loss": 1.1837,
+      "step": 21280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8416530839185345e-05,
+      "loss": 1.2721,
+      "step": 21290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841576918975734e-05,
+      "loss": 1.3592,
+      "step": 21300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841500754032934e-05,
+      "loss": 1.2867,
+      "step": 21310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841424589090134e-05,
+      "loss": 1.1665,
+      "step": 21320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841348424147334e-05,
+      "loss": 1.281,
+      "step": 21330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.841272259204533e-05,
+      "loss": 1.2096,
+      "step": 21340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8411960942617336e-05,
+      "loss": 1.2639,
+      "step": 21350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8411199293189334e-05,
+      "loss": 1.2172,
+      "step": 21360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.841043764376133e-05,
+      "loss": 1.1419,
+      "step": 21370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840967599433333e-05,
+      "loss": 1.2104,
+      "step": 21380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840891434490533e-05,
+      "loss": 1.3131,
+      "step": 21390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8408152695477324e-05,
+      "loss": 1.2956,
+      "step": 21400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840739104604933e-05,
+      "loss": 1.2397,
+      "step": 21410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8406629396621325e-05,
+      "loss": 1.3306,
+      "step": 21420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840586774719332e-05,
+      "loss": 1.2999,
+      "step": 21430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840510609776533e-05,
+      "loss": 1.2239,
+      "step": 21440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8404344448337324e-05,
+      "loss": 1.2699,
+      "step": 21450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840358279890932e-05,
+      "loss": 1.3172,
+      "step": 21460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840282114948132e-05,
+      "loss": 1.25,
+      "step": 21470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840205950005332e-05,
+      "loss": 1.1862,
+      "step": 21480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8401297850625314e-05,
+      "loss": 1.2449,
+      "step": 21490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.840053620119732e-05,
+      "loss": 1.261,
+      "step": 21500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.839977455176931e-05,
+      "loss": 1.1279,
+      "step": 21510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8399012902341313e-05,
+      "loss": 1.1863,
+      "step": 21520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.839825125291331e-05,
+      "loss": 1.2709,
+      "step": 21530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.839748960348531e-05,
+      "loss": 1.2581,
+      "step": 21540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8396727954057306e-05,
+      "loss": 1.2406,
+      "step": 21550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.839596630462931e-05,
+      "loss": 1.252,
+      "step": 21560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.83952046552013e-05,
+      "loss": 1.2879,
+      "step": 21570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8394443005773305e-05,
+      "loss": 1.1843,
+      "step": 21580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83936813563453e-05,
+      "loss": 1.1222,
+      "step": 21590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83929197069173e-05,
+      "loss": 1.1581,
+      "step": 21600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83921580574893e-05,
+      "loss": 1.288,
+      "step": 21610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83913964080613e-05,
+      "loss": 1.3583,
+      "step": 21620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83906347586333e-05,
+      "loss": 1.2866,
+      "step": 21630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83898731092053e-05,
+      "loss": 1.1991,
+      "step": 21640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83891114597773e-05,
+      "loss": 1.1628,
+      "step": 21650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838834981034929e-05,
+      "loss": 1.3327,
+      "step": 21660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8387588160921296e-05,
+      "loss": 1.194,
+      "step": 21670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838682651149329e-05,
+      "loss": 1.2817,
+      "step": 21680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838606486206529e-05,
+      "loss": 1.2074,
+      "step": 21690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838530321263729e-05,
+      "loss": 1.2835,
+      "step": 21700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838454156320929e-05,
+      "loss": 1.2881,
+      "step": 21710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838377991378128e-05,
+      "loss": 1.2756,
+      "step": 21720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838301826435329e-05,
+      "loss": 1.3132,
+      "step": 21730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8382256614925285e-05,
+      "loss": 1.1344,
+      "step": 21740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838149496549728e-05,
+      "loss": 1.4009,
+      "step": 21750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.838073331606928e-05,
+      "loss": 1.2472,
+      "step": 21760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8379971666641284e-05,
+      "loss": 1.2147,
+      "step": 21770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8379210017213275e-05,
+      "loss": 1.3306,
+      "step": 21780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.837844836778528e-05,
+      "loss": 1.3509,
+      "step": 21790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8377686718357276e-05,
+      "loss": 1.2202,
+      "step": 21800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8376925068929274e-05,
+      "loss": 1.323,
+      "step": 21810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.837616341950127e-05,
+      "loss": 1.1532,
+      "step": 21820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8375401770073276e-05,
+      "loss": 1.2539,
+      "step": 21830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.837464012064527e-05,
+      "loss": 1.4802,
+      "step": 21840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.837387847121727e-05,
+      "loss": 1.3336,
+      "step": 21850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8373116821789275e-05,
+      "loss": 1.3275,
+      "step": 21860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8372355172361265e-05,
+      "loss": 1.3051,
+      "step": 21870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.837159352293327e-05,
+      "loss": 1.2765,
+      "step": 21880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.837083187350527e-05,
+      "loss": 1.2368,
+      "step": 21890
+    },
+    {
+      "epoch": 1.0,
+      "eval_cer": 0.6322929220178798,
+      "eval_em": 0.007379375591296121,
+      "eval_f1": 0.007379375591296121,
+      "eval_loss": 1.0779038667678833,
+      "eval_runtime": 1007.4863,
+      "eval_samples_per_second": 10.491,
+      "eval_steps_per_second": 1.312,
+      "eval_wer": 0.9926206244087039,
+      "step": 21899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8370070224077265e-05,
+      "loss": 1.1854,
+      "step": 21900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836930857464926e-05,
+      "loss": 1.2878,
+      "step": 21910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8368546925221266e-05,
+      "loss": 1.245,
+      "step": 21920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836778527579326e-05,
+      "loss": 1.1533,
+      "step": 21930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836702362636526e-05,
+      "loss": 1.2864,
+      "step": 21940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836626197693725e-05,
+      "loss": 1.2006,
+      "step": 21950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8365500327509256e-05,
+      "loss": 1.1536,
+      "step": 21960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8364738678081254e-05,
+      "loss": 1.323,
+      "step": 21970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836397702865325e-05,
+      "loss": 1.2002,
+      "step": 21980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836321537922525e-05,
+      "loss": 1.3121,
+      "step": 21990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.836245372979725e-05,
+      "loss": 1.291,
+      "step": 22000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.836169208036925e-05,
+      "loss": 1.2181,
+      "step": 22010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.836093043094125e-05,
+      "loss": 1.274,
+      "step": 22020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.836016878151325e-05,
+      "loss": 1.2819,
+      "step": 22030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835940713208524e-05,
+      "loss": 1.0689,
+      "step": 22040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835864548265725e-05,
+      "loss": 1.0957,
+      "step": 22050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8357883833229244e-05,
+      "loss": 1.2621,
+      "step": 22060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835712218380124e-05,
+      "loss": 1.1641,
+      "step": 22070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835636053437324e-05,
+      "loss": 1.2616,
+      "step": 22080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8355598884945244e-05,
+      "loss": 1.2359,
+      "step": 22090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8354837235517234e-05,
+      "loss": 1.2527,
+      "step": 22100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835407558608924e-05,
+      "loss": 1.2088,
+      "step": 22110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8353313936661236e-05,
+      "loss": 1.2296,
+      "step": 22120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8352552287233233e-05,
+      "loss": 1.1857,
+      "step": 22130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835179063780523e-05,
+      "loss": 1.2708,
+      "step": 22140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8351028988377235e-05,
+      "loss": 1.2729,
+      "step": 22150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8350267338949226e-05,
+      "loss": 1.202,
+      "step": 22160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834950568952123e-05,
+      "loss": 1.1265,
+      "step": 22170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834874404009323e-05,
+      "loss": 1.1445,
+      "step": 22180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8347982390665225e-05,
+      "loss": 1.2163,
+      "step": 22190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834722074123722e-05,
+      "loss": 1.104,
+      "step": 22200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834645909180923e-05,
+      "loss": 1.3014,
+      "step": 22210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8345697442381224e-05,
+      "loss": 1.1224,
+      "step": 22220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.834493579295322e-05,
+      "loss": 1.2905,
+      "step": 22230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8344174143525226e-05,
+      "loss": 1.2639,
+      "step": 22240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.834341249409722e-05,
+      "loss": 1.2938,
+      "step": 22250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.834265084466922e-05,
+      "loss": 1.1603,
+      "step": 22260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.834188919524122e-05,
+      "loss": 1.3096,
+      "step": 22270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8341127545813216e-05,
+      "loss": 1.1044,
+      "step": 22280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.834036589638521e-05,
+      "loss": 1.1196,
+      "step": 22290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.833960424695722e-05,
+      "loss": 1.2766,
+      "step": 22300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.833884259752921e-05,
+      "loss": 1.1448,
+      "step": 22310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.833808094810121e-05,
+      "loss": 1.257,
+      "step": 22320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.833731929867321e-05,
+      "loss": 1.3964,
+      "step": 22330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.833655764924521e-05,
+      "loss": 1.2438,
+      "step": 22340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8335795999817205e-05,
+      "loss": 1.221,
+      "step": 22350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.833503435038921e-05,
+      "loss": 1.2012,
+      "step": 22360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.83342727009612e-05,
+      "loss": 1.1243,
+      "step": 22370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8333511051533204e-05,
+      "loss": 1.2954,
+      "step": 22380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.83327494021052e-05,
+      "loss": 1.2852,
+      "step": 22390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.83319877526772e-05,
+      "loss": 1.2934,
+      "step": 22400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8331226103249196e-05,
+      "loss": 1.3162,
+      "step": 22410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8330464453821194e-05,
+      "loss": 1.2582,
+      "step": 22420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.83297028043932e-05,
+      "loss": 1.0741,
+      "step": 22430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8328941154965196e-05,
+      "loss": 1.2019,
+      "step": 22440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832817950553719e-05,
+      "loss": 1.1964,
+      "step": 22450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832741785610919e-05,
+      "loss": 1.2421,
+      "step": 22460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8326656206681195e-05,
+      "loss": 1.2224,
+      "step": 22470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8325894557253185e-05,
+      "loss": 1.2123,
+      "step": 22480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832513290782519e-05,
+      "loss": 1.1486,
+      "step": 22490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832437125839719e-05,
+      "loss": 1.1483,
+      "step": 22500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8323609608969185e-05,
+      "loss": 1.0734,
+      "step": 22510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832284795954118e-05,
+      "loss": 1.1572,
+      "step": 22520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8322086310113186e-05,
+      "loss": 1.0695,
+      "step": 22530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832132466068518e-05,
+      "loss": 1.2142,
+      "step": 22540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.832056301125718e-05,
+      "loss": 1.1478,
+      "step": 22550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831980136182918e-05,
+      "loss": 1.2305,
+      "step": 22560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8319039712401176e-05,
+      "loss": 1.266,
+      "step": 22570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8318278062973174e-05,
+      "loss": 1.3058,
+      "step": 22580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831751641354518e-05,
+      "loss": 1.1244,
+      "step": 22590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8316754764117175e-05,
+      "loss": 1.3456,
+      "step": 22600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831599311468917e-05,
+      "loss": 1.2535,
+      "step": 22610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831523146526117e-05,
+      "loss": 1.1486,
+      "step": 22620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831446981583317e-05,
+      "loss": 1.198,
+      "step": 22630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831370816640517e-05,
+      "loss": 1.4,
+      "step": 22640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831294651697717e-05,
+      "loss": 1.2613,
+      "step": 22650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.831218486754917e-05,
+      "loss": 1.2188,
+      "step": 22660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8311423218121164e-05,
+      "loss": 1.2338,
+      "step": 22670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.831066156869317e-05,
+      "loss": 1.2202,
+      "step": 22680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830989991926516e-05,
+      "loss": 1.1539,
+      "step": 22690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8309138269837164e-05,
+      "loss": 1.2038,
+      "step": 22700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830837662040916e-05,
+      "loss": 1.0197,
+      "step": 22710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830761497098116e-05,
+      "loss": 1.1498,
+      "step": 22720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8306853321553156e-05,
+      "loss": 1.2293,
+      "step": 22730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830609167212516e-05,
+      "loss": 1.1303,
+      "step": 22740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830533002269715e-05,
+      "loss": 1.2529,
+      "step": 22750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8304568373269155e-05,
+      "loss": 1.1052,
+      "step": 22760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830380672384115e-05,
+      "loss": 1.149,
+      "step": 22770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830304507441315e-05,
+      "loss": 1.1269,
+      "step": 22780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830228342498515e-05,
+      "loss": 1.112,
+      "step": 22790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830152177555715e-05,
+      "loss": 1.2063,
+      "step": 22800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.830076012612915e-05,
+      "loss": 1.2648,
+      "step": 22810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.829999847670115e-05,
+      "loss": 1.214,
+      "step": 22820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.829923682727315e-05,
+      "loss": 1.3718,
+      "step": 22830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.829847517784514e-05,
+      "loss": 1.2929,
+      "step": 22840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8297713528417146e-05,
+      "loss": 1.2981,
+      "step": 22850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.829695187898914e-05,
+      "loss": 1.2345,
+      "step": 22860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.829619022956114e-05,
+      "loss": 1.1914,
+      "step": 22870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.829542858013314e-05,
+      "loss": 1.2037,
+      "step": 22880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8294666930705136e-05,
+      "loss": 1.1815,
+      "step": 22890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.829390528127713e-05,
+      "loss": 1.3148,
+      "step": 22900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.829314363184914e-05,
+      "loss": 1.2793,
+      "step": 22910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.829238198242113e-05,
+      "loss": 1.1982,
+      "step": 22920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.829162033299313e-05,
+      "loss": 1.3043,
+      "step": 22930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.829085868356513e-05,
+      "loss": 1.1372,
+      "step": 22940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.829009703413713e-05,
+      "loss": 1.1911,
+      "step": 22950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8289335384709125e-05,
+      "loss": 1.2262,
+      "step": 22960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828857373528113e-05,
+      "loss": 1.1537,
+      "step": 22970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8287812085853127e-05,
+      "loss": 1.1619,
+      "step": 22980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8287050436425124e-05,
+      "loss": 1.2492,
+      "step": 22990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828628878699712e-05,
+      "loss": 1.3632,
+      "step": 23000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828552713756912e-05,
+      "loss": 1.1402,
+      "step": 23010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828476548814112e-05,
+      "loss": 1.1748,
+      "step": 23020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828400383871312e-05,
+      "loss": 1.2785,
+      "step": 23030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828324218928512e-05,
+      "loss": 1.2996,
+      "step": 23040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8282480539857116e-05,
+      "loss": 1.1412,
+      "step": 23050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828171889042912e-05,
+      "loss": 1.2782,
+      "step": 23060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.828095724100111e-05,
+      "loss": 1.2051,
+      "step": 23070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8280195591573115e-05,
+      "loss": 1.147,
+      "step": 23080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.827943394214511e-05,
+      "loss": 1.119,
+      "step": 23090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.827867229271711e-05,
+      "loss": 1.279,
+      "step": 23100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.827791064328911e-05,
+      "loss": 1.2762,
+      "step": 23110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.827714899386111e-05,
+      "loss": 1.2601,
+      "step": 23120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82763873444331e-05,
+      "loss": 1.3323,
+      "step": 23130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8275625695005106e-05,
+      "loss": 1.4022,
+      "step": 23140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8274864045577104e-05,
+      "loss": 1.1355,
+      "step": 23150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82741023961491e-05,
+      "loss": 1.2101,
+      "step": 23160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82733407467211e-05,
+      "loss": 1.2106,
+      "step": 23170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82725790972931e-05,
+      "loss": 1.3174,
+      "step": 23180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82718174478651e-05,
+      "loss": 1.2074,
+      "step": 23190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82710557984371e-05,
+      "loss": 1.2112,
+      "step": 23200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8270294149009095e-05,
+      "loss": 1.1929,
+      "step": 23210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.826953249958109e-05,
+      "loss": 1.2831,
+      "step": 23220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82687708501531e-05,
+      "loss": 1.3465,
+      "step": 23230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8268009200725095e-05,
+      "loss": 1.1884,
+      "step": 23240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.826724755129709e-05,
+      "loss": 1.1076,
+      "step": 23250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.826648590186909e-05,
+      "loss": 1.1964,
+      "step": 23260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8265724252441094e-05,
+      "loss": 1.1482,
+      "step": 23270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8264962603013084e-05,
+      "loss": 1.2461,
+      "step": 23280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.826420095358509e-05,
+      "loss": 1.2074,
+      "step": 23290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8263439304157086e-05,
+      "loss": 1.1335,
+      "step": 23300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8262677654729084e-05,
+      "loss": 1.2734,
+      "step": 23310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.826191600530108e-05,
+      "loss": 1.2692,
+      "step": 23320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.826115435587308e-05,
+      "loss": 1.194,
+      "step": 23330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8260392706445076e-05,
+      "loss": 1.2288,
+      "step": 23340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825963105701708e-05,
+      "loss": 1.3369,
+      "step": 23350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825886940758908e-05,
+      "loss": 1.1638,
+      "step": 23360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8258107758161075e-05,
+      "loss": 1.2361,
+      "step": 23370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825734610873307e-05,
+      "loss": 1.2276,
+      "step": 23380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825658445930507e-05,
+      "loss": 1.2223,
+      "step": 23390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8255822809877074e-05,
+      "loss": 1.1419,
+      "step": 23400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825506116044907e-05,
+      "loss": 1.2746,
+      "step": 23410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825429951102107e-05,
+      "loss": 1.152,
+      "step": 23420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825353786159307e-05,
+      "loss": 1.1152,
+      "step": 23430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825277621216507e-05,
+      "loss": 1.1372,
+      "step": 23440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.825201456273706e-05,
+      "loss": 1.2421,
+      "step": 23450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8251252913309066e-05,
+      "loss": 1.0743,
+      "step": 23460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8250491263881063e-05,
+      "loss": 1.2517,
+      "step": 23470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.824972961445306e-05,
+      "loss": 1.1721,
+      "step": 23480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.824896796502506e-05,
+      "loss": 1.1596,
+      "step": 23490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.824820631559706e-05,
+      "loss": 1.1796,
+      "step": 23500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.824744466616905e-05,
+      "loss": 1.2366,
+      "step": 23510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.824668301674106e-05,
+      "loss": 1.1821,
+      "step": 23520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8245921367313055e-05,
+      "loss": 1.1646,
+      "step": 23530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.824515971788505e-05,
+      "loss": 1.2256,
+      "step": 23540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.824439806845705e-05,
+      "loss": 1.2433,
+      "step": 23550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8243636419029054e-05,
+      "loss": 1.2689,
+      "step": 23560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.824287476960105e-05,
+      "loss": 1.1175,
+      "step": 23570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.824211312017305e-05,
+      "loss": 1.2547,
+      "step": 23580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8241351470745047e-05,
+      "loss": 1.2134,
+      "step": 23590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8240589821317044e-05,
+      "loss": 1.1595,
+      "step": 23600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823982817188905e-05,
+      "loss": 1.2575,
+      "step": 23610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8239066522461046e-05,
+      "loss": 1.0051,
+      "step": 23620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823830487303304e-05,
+      "loss": 1.3256,
+      "step": 23630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823754322360504e-05,
+      "loss": 1.1372,
+      "step": 23640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8236781574177045e-05,
+      "loss": 1.1456,
+      "step": 23650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8236019924749036e-05,
+      "loss": 1.122,
+      "step": 23660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823525827532104e-05,
+      "loss": 1.3343,
+      "step": 23670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823449662589304e-05,
+      "loss": 1.2944,
+      "step": 23680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8233734976465035e-05,
+      "loss": 1.2366,
+      "step": 23690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823297332703703e-05,
+      "loss": 1.2301,
+      "step": 23700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8232211677609037e-05,
+      "loss": 1.0919,
+      "step": 23710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823145002818103e-05,
+      "loss": 1.1815,
+      "step": 23720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.823068837875303e-05,
+      "loss": 1.3229,
+      "step": 23730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.822992672932503e-05,
+      "loss": 1.1632,
+      "step": 23740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8229165079897026e-05,
+      "loss": 1.1956,
+      "step": 23750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8228403430469024e-05,
+      "loss": 1.2027,
+      "step": 23760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822764178104103e-05,
+      "loss": 1.125,
+      "step": 23770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8226880131613026e-05,
+      "loss": 1.1119,
+      "step": 23780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822611848218502e-05,
+      "loss": 1.2216,
+      "step": 23790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822535683275702e-05,
+      "loss": 1.165,
+      "step": 23800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822459518332902e-05,
+      "loss": 1.2893,
+      "step": 23810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822383353390102e-05,
+      "loss": 1.2738,
+      "step": 23820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822307188447301e-05,
+      "loss": 1.2322,
+      "step": 23830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822231023504502e-05,
+      "loss": 1.1561,
+      "step": 23840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8221548585617015e-05,
+      "loss": 1.1573,
+      "step": 23850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822078693618901e-05,
+      "loss": 1.1496,
+      "step": 23860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.822002528676101e-05,
+      "loss": 1.2307,
+      "step": 23870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8219263637333014e-05,
+      "loss": 1.2255,
+      "step": 23880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8218501987905004e-05,
+      "loss": 1.2478,
+      "step": 23890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.821774033847701e-05,
+      "loss": 1.3304,
+      "step": 23900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8216978689049006e-05,
+      "loss": 1.1051,
+      "step": 23910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8216217039621004e-05,
+      "loss": 1.113,
+      "step": 23920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8215455390193e-05,
+      "loss": 1.2474,
+      "step": 23930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8214693740765005e-05,
+      "loss": 1.2409,
+      "step": 23940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8213932091337e-05,
+      "loss": 1.2772,
+      "step": 23950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8213170441909e-05,
+      "loss": 1.0788,
+      "step": 23960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8212408792481e-05,
+      "loss": 1.2892,
+      "step": 23970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8211647143052995e-05,
+      "loss": 1.1256,
+      "step": 23980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8210885493625e-05,
+      "loss": 1.2677,
+      "step": 23990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8210123844197e-05,
+      "loss": 1.2278,
+      "step": 24000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8209362194768994e-05,
+      "loss": 1.2787,
+      "step": 24010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820860054534099e-05,
+      "loss": 1.0705,
+      "step": 24020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8207838895912996e-05,
+      "loss": 1.1525,
+      "step": 24030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820707724648499e-05,
+      "loss": 1.289,
+      "step": 24040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820631559705699e-05,
+      "loss": 1.2312,
+      "step": 24050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820555394762899e-05,
+      "loss": 1.366,
+      "step": 24060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8204792298200986e-05,
+      "loss": 1.3197,
+      "step": 24070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8204030648772983e-05,
+      "loss": 1.3334,
+      "step": 24080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820326899934499e-05,
+      "loss": 1.1672,
+      "step": 24090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820250734991698e-05,
+      "loss": 1.2506,
+      "step": 24100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820174570048898e-05,
+      "loss": 1.2112,
+      "step": 24110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820098405106098e-05,
+      "loss": 1.1678,
+      "step": 24120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.820022240163298e-05,
+      "loss": 1.1666,
+      "step": 24130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8199460752204975e-05,
+      "loss": 1.1297,
+      "step": 24140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.819869910277698e-05,
+      "loss": 1.1229,
+      "step": 24150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.819793745334898e-05,
+      "loss": 1.2129,
+      "step": 24160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8197175803920974e-05,
+      "loss": 1.1094,
+      "step": 24170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.819641415449297e-05,
+      "loss": 1.2102,
+      "step": 24180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.819565250506497e-05,
+      "loss": 1.3632,
+      "step": 24190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.819489085563697e-05,
+      "loss": 1.2586,
+      "step": 24200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.819412920620897e-05,
+      "loss": 1.1766,
+      "step": 24210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.819336755678097e-05,
+      "loss": 1.2283,
+      "step": 24220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8192605907352966e-05,
+      "loss": 1.195,
+      "step": 24230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.819184425792497e-05,
+      "loss": 1.0962,
+      "step": 24240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.819108260849696e-05,
+      "loss": 1.106,
+      "step": 24250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8190320959068965e-05,
+      "loss": 1.2049,
+      "step": 24260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8189559309640956e-05,
+      "loss": 1.3132,
+      "step": 24270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818879766021296e-05,
+      "loss": 1.2084,
+      "step": 24280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818803601078496e-05,
+      "loss": 1.2368,
+      "step": 24290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8187274361356955e-05,
+      "loss": 1.0407,
+      "step": 24300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818651271192895e-05,
+      "loss": 1.2158,
+      "step": 24310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8185751062500957e-05,
+      "loss": 1.2811,
+      "step": 24320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8184989413072954e-05,
+      "loss": 1.1647,
+      "step": 24330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818422776364495e-05,
+      "loss": 1.0941,
+      "step": 24340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818346611421695e-05,
+      "loss": 1.1081,
+      "step": 24350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8182704464788946e-05,
+      "loss": 1.2726,
+      "step": 24360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818194281536095e-05,
+      "loss": 1.2595,
+      "step": 24370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.818118116593295e-05,
+      "loss": 1.2744,
+      "step": 24380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8180419516504946e-05,
+      "loss": 1.0562,
+      "step": 24390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.817965786707694e-05,
+      "loss": 1.2587,
+      "step": 24400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.817889621764895e-05,
+      "loss": 1.1776,
+      "step": 24410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817813456822094e-05,
+      "loss": 1.2443,
+      "step": 24420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817737291879294e-05,
+      "loss": 1.1192,
+      "step": 24430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817661126936494e-05,
+      "loss": 1.1896,
+      "step": 24440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817584961993694e-05,
+      "loss": 1.2045,
+      "step": 24450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8175087970508935e-05,
+      "loss": 1.2033,
+      "step": 24460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817432632108094e-05,
+      "loss": 1.4437,
+      "step": 24470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817356467165293e-05,
+      "loss": 1.1611,
+      "step": 24480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8172803022224934e-05,
+      "loss": 1.1923,
+      "step": 24490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817204137279693e-05,
+      "loss": 1.1125,
+      "step": 24500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817127972336893e-05,
+      "loss": 1.1752,
+      "step": 24510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8170518073940926e-05,
+      "loss": 1.29,
+      "step": 24520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816975642451293e-05,
+      "loss": 1.2481,
+      "step": 24530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816899477508493e-05,
+      "loss": 1.2704,
+      "step": 24540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8168233125656925e-05,
+      "loss": 1.2776,
+      "step": 24550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816747147622892e-05,
+      "loss": 1.1438,
+      "step": 24560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816670982680092e-05,
+      "loss": 1.1422,
+      "step": 24570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8165948177372925e-05,
+      "loss": 1.1689,
+      "step": 24580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816518652794492e-05,
+      "loss": 1.2219,
+      "step": 24590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816442487851692e-05,
+      "loss": 1.1753,
+      "step": 24600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816366322908892e-05,
+      "loss": 1.2829,
+      "step": 24610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816290157966092e-05,
+      "loss": 1.2559,
+      "step": 24620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816213993023291e-05,
+      "loss": 1.1809,
+      "step": 24630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8161378280804916e-05,
+      "loss": 1.1981,
+      "step": 24640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8160616631376914e-05,
+      "loss": 1.3037,
+      "step": 24650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815985498194891e-05,
+      "loss": 1.1985,
+      "step": 24660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815909333252091e-05,
+      "loss": 1.2929,
+      "step": 24670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815833168309291e-05,
+      "loss": 1.163,
+      "step": 24680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8157570033664903e-05,
+      "loss": 1.3794,
+      "step": 24690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815680838423691e-05,
+      "loss": 1.1883,
+      "step": 24700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81560467348089e-05,
+      "loss": 1.338,
+      "step": 24710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81552850853809e-05,
+      "loss": 1.4828,
+      "step": 24720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81545234359529e-05,
+      "loss": 1.1823,
+      "step": 24730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81537617865249e-05,
+      "loss": 1.2514,
+      "step": 24740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81530001370969e-05,
+      "loss": 1.4326,
+      "step": 24750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81522384876689e-05,
+      "loss": 1.2406,
+      "step": 24760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81514768382409e-05,
+      "loss": 1.172,
+      "step": 24770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8150715188812894e-05,
+      "loss": 1.2535,
+      "step": 24780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81499535393849e-05,
+      "loss": 1.0695,
+      "step": 24790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814919188995689e-05,
+      "loss": 1.2838,
+      "step": 24800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8148430240528893e-05,
+      "loss": 1.2518,
+      "step": 24810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814766859110089e-05,
+      "loss": 1.1289,
+      "step": 24820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814690694167289e-05,
+      "loss": 1.2218,
+      "step": 24830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8146145292244886e-05,
+      "loss": 1.1349,
+      "step": 24840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814538364281689e-05,
+      "loss": 1.0187,
+      "step": 24850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.814462199338888e-05,
+      "loss": 1.2239,
+      "step": 24860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8143860343960885e-05,
+      "loss": 1.313,
+      "step": 24870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.814309869453288e-05,
+      "loss": 1.0691,
+      "step": 24880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.814233704510488e-05,
+      "loss": 1.2654,
+      "step": 24890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.814157539567688e-05,
+      "loss": 1.4193,
+      "step": 24900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.814081374624888e-05,
+      "loss": 1.1268,
+      "step": 24910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.814005209682088e-05,
+      "loss": 1.1744,
+      "step": 24920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8139290447392877e-05,
+      "loss": 1.4263,
+      "step": 24930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8138528797964874e-05,
+      "loss": 1.1042,
+      "step": 24940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813776714853687e-05,
+      "loss": 1.2803,
+      "step": 24950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8137005499108876e-05,
+      "loss": 1.2441,
+      "step": 24960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813624384968087e-05,
+      "loss": 1.1235,
+      "step": 24970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813548220025287e-05,
+      "loss": 1.2233,
+      "step": 24980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813472055082487e-05,
+      "loss": 1.0849,
+      "step": 24990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813395890139687e-05,
+      "loss": 1.331,
+      "step": 25000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813319725196886e-05,
+      "loss": 1.2209,
+      "step": 25010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813243560254087e-05,
+      "loss": 1.1902,
+      "step": 25020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8131673953112865e-05,
+      "loss": 1.284,
+      "step": 25030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813091230368486e-05,
+      "loss": 1.2964,
+      "step": 25040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.813015065425686e-05,
+      "loss": 1.2722,
+      "step": 25050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8129389004828864e-05,
+      "loss": 1.2148,
+      "step": 25060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8128627355400855e-05,
+      "loss": 1.3041,
+      "step": 25070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812786570597286e-05,
+      "loss": 1.3774,
+      "step": 25080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8127104056544856e-05,
+      "loss": 1.2457,
+      "step": 25090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8126342407116854e-05,
+      "loss": 1.1727,
+      "step": 25100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812558075768885e-05,
+      "loss": 1.2226,
+      "step": 25110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8124819108260856e-05,
+      "loss": 1.1841,
+      "step": 25120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812405745883285e-05,
+      "loss": 1.2216,
+      "step": 25130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812329580940485e-05,
+      "loss": 1.3558,
+      "step": 25140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812253415997685e-05,
+      "loss": 1.2272,
+      "step": 25150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8121772510548845e-05,
+      "loss": 1.1688,
+      "step": 25160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812101086112085e-05,
+      "loss": 1.4047,
+      "step": 25170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.812024921169284e-05,
+      "loss": 1.232,
+      "step": 25180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8119487562264845e-05,
+      "loss": 1.1951,
+      "step": 25190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811872591283684e-05,
+      "loss": 1.1195,
+      "step": 25200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811796426340884e-05,
+      "loss": 1.2386,
+      "step": 25210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811720261398084e-05,
+      "loss": 1.2216,
+      "step": 25220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811644096455284e-05,
+      "loss": 1.1838,
+      "step": 25230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811567931512483e-05,
+      "loss": 1.2526,
+      "step": 25240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8114917665696836e-05,
+      "loss": 1.3516,
+      "step": 25250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8114156016268834e-05,
+      "loss": 1.177,
+      "step": 25260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811339436684083e-05,
+      "loss": 1.1517,
+      "step": 25270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811263271741283e-05,
+      "loss": 1.2544,
+      "step": 25280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.811187106798483e-05,
+      "loss": 1.1322,
+      "step": 25290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8111109418556823e-05,
+      "loss": 1.1772,
+      "step": 25300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.811034776912883e-05,
+      "loss": 1.225,
+      "step": 25310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8109586119700825e-05,
+      "loss": 1.3109,
+      "step": 25320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.810882447027282e-05,
+      "loss": 1.1737,
+      "step": 25330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.810806282084483e-05,
+      "loss": 1.1632,
+      "step": 25340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8107301171416824e-05,
+      "loss": 1.1669,
+      "step": 25350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.810653952198882e-05,
+      "loss": 1.0577,
+      "step": 25360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.810577787256082e-05,
+      "loss": 1.0883,
+      "step": 25370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8105016223132824e-05,
+      "loss": 1.2796,
+      "step": 25380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8104254573704814e-05,
+      "loss": 1.0785,
+      "step": 25390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.810349292427682e-05,
+      "loss": 1.2666,
+      "step": 25400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8102731274848816e-05,
+      "loss": 1.1997,
+      "step": 25410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8101969625420813e-05,
+      "loss": 1.259,
+      "step": 25420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.810120797599281e-05,
+      "loss": 1.1608,
+      "step": 25430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8100446326564815e-05,
+      "loss": 1.3033,
+      "step": 25440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8099684677136806e-05,
+      "loss": 1.2808,
+      "step": 25450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.809892302770881e-05,
+      "loss": 1.2867,
+      "step": 25460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.809816137828081e-05,
+      "loss": 1.2044,
+      "step": 25470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8097399728852805e-05,
+      "loss": 1.2552,
+      "step": 25480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.80966380794248e-05,
+      "loss": 1.3464,
+      "step": 25490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.809587642999681e-05,
+      "loss": 1.3651,
+      "step": 25500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.80951147805688e-05,
+      "loss": 1.1696,
+      "step": 25510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.80943531311408e-05,
+      "loss": 1.2168,
+      "step": 25520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.80935914817128e-05,
+      "loss": 1.2745,
+      "step": 25530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8092829832284797e-05,
+      "loss": 1.1716,
+      "step": 25540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.80920681828568e-05,
+      "loss": 1.2431,
+      "step": 25550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.80913065334288e-05,
+      "loss": 1.3064,
+      "step": 25560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8090544884000796e-05,
+      "loss": 1.2384,
+      "step": 25570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808978323457279e-05,
+      "loss": 1.3801,
+      "step": 25580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.80890215851448e-05,
+      "loss": 1.1047,
+      "step": 25590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808825993571679e-05,
+      "loss": 1.0679,
+      "step": 25600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808749828628879e-05,
+      "loss": 1.2307,
+      "step": 25610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808673663686079e-05,
+      "loss": 1.2054,
+      "step": 25620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808597498743279e-05,
+      "loss": 1.1889,
+      "step": 25630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8085213338004785e-05,
+      "loss": 1.1461,
+      "step": 25640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808445168857678e-05,
+      "loss": 1.227,
+      "step": 25650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808369003914878e-05,
+      "loss": 1.1526,
+      "step": 25660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8082928389720784e-05,
+      "loss": 1.1832,
+      "step": 25670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8082166740292775e-05,
+      "loss": 1.2764,
+      "step": 25680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.808140509086478e-05,
+      "loss": 1.1497,
+      "step": 25690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8080643441436776e-05,
+      "loss": 1.2151,
+      "step": 25700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8079881792008774e-05,
+      "loss": 1.2388,
+      "step": 25710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.807912014258078e-05,
+      "loss": 1.2168,
+      "step": 25720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8078358493152776e-05,
+      "loss": 1.1431,
+      "step": 25730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.807759684372477e-05,
+      "loss": 1.2833,
+      "step": 25740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.807683519429677e-05,
+      "loss": 1.1798,
+      "step": 25750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8076073544868775e-05,
+      "loss": 1.2222,
+      "step": 25760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8075311895440765e-05,
+      "loss": 1.3098,
+      "step": 25770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.807455024601277e-05,
+      "loss": 1.1531,
+      "step": 25780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.807378859658477e-05,
+      "loss": 1.3752,
+      "step": 25790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8073026947156765e-05,
+      "loss": 1.2191,
+      "step": 25800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.807226529772876e-05,
+      "loss": 1.0937,
+      "step": 25810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8071503648300766e-05,
+      "loss": 1.1788,
+      "step": 25820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.807074199887276e-05,
+      "loss": 1.1971,
+      "step": 25830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806998034944476e-05,
+      "loss": 1.0933,
+      "step": 25840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806921870001676e-05,
+      "loss": 1.3002,
+      "step": 25850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8068457050588756e-05,
+      "loss": 1.2735,
+      "step": 25860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8067695401160754e-05,
+      "loss": 1.3464,
+      "step": 25870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806693375173276e-05,
+      "loss": 1.2591,
+      "step": 25880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806617210230475e-05,
+      "loss": 1.353,
+      "step": 25890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806541045287675e-05,
+      "loss": 1.1614,
+      "step": 25900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806464880344875e-05,
+      "loss": 1.1797,
+      "step": 25910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806388715402075e-05,
+      "loss": 1.313,
+      "step": 25920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806312550459275e-05,
+      "loss": 1.1017,
+      "step": 25930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806236385516475e-05,
+      "loss": 1.2453,
+      "step": 25940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.806160220573675e-05,
+      "loss": 1.0768,
+      "step": 25950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8060840556308744e-05,
+      "loss": 1.1932,
+      "step": 25960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.806007890688075e-05,
+      "loss": 1.073,
+      "step": 25970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805931725745274e-05,
+      "loss": 1.2219,
+      "step": 25980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8058555608024744e-05,
+      "loss": 1.1695,
+      "step": 25990
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805779395859674e-05,
+      "loss": 1.2288,
+      "step": 26000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805703230916874e-05,
+      "loss": 1.2101,
+      "step": 26010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8056270659740736e-05,
+      "loss": 1.1864,
+      "step": 26020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805550901031274e-05,
+      "loss": 1.2154,
+      "step": 26030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805474736088473e-05,
+      "loss": 1.4491,
+      "step": 26040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8053985711456735e-05,
+      "loss": 1.0808,
+      "step": 26050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805322406202873e-05,
+      "loss": 1.1818,
+      "step": 26060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805246241260073e-05,
+      "loss": 1.2787,
+      "step": 26070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805170076317273e-05,
+      "loss": 1.1244,
+      "step": 26080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8050939113744725e-05,
+      "loss": 1.1938,
+      "step": 26090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.805017746431672e-05,
+      "loss": 1.1573,
+      "step": 26100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.804941581488873e-05,
+      "loss": 1.1613,
+      "step": 26110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8048654165460724e-05,
+      "loss": 1.1301,
+      "step": 26120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.804789251603272e-05,
+      "loss": 1.0868,
+      "step": 26130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8047130866604726e-05,
+      "loss": 1.3125,
+      "step": 26140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8046369217176717e-05,
+      "loss": 1.1533,
+      "step": 26150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.804560756774872e-05,
+      "loss": 1.1555,
+      "step": 26160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.804484591832072e-05,
+      "loss": 1.2236,
+      "step": 26170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8044084268892716e-05,
+      "loss": 1.271,
+      "step": 26180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.804332261946471e-05,
+      "loss": 1.0853,
+      "step": 26190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.804256097003672e-05,
+      "loss": 1.2701,
+      "step": 26200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.804179932060871e-05,
+      "loss": 1.2027,
+      "step": 26210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.804103767118071e-05,
+      "loss": 1.2814,
+      "step": 26220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.804027602175271e-05,
+      "loss": 1.2504,
+      "step": 26230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.803951437232471e-05,
+      "loss": 1.2291,
+      "step": 26240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8038752722896705e-05,
+      "loss": 1.2056,
+      "step": 26250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.803799107346871e-05,
+      "loss": 1.2247,
+      "step": 26260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80372294240407e-05,
+      "loss": 1.1011,
+      "step": 26270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8036467774612704e-05,
+      "loss": 1.2278,
+      "step": 26280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80357061251847e-05,
+      "loss": 1.2677,
+      "step": 26290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80349444757567e-05,
+      "loss": 1.1231,
+      "step": 26300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8034182826328696e-05,
+      "loss": 1.0831,
+      "step": 26310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80334211769007e-05,
+      "loss": 1.3118,
+      "step": 26320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80326595274727e-05,
+      "loss": 1.2114,
+      "step": 26330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8031897878044696e-05,
+      "loss": 1.1145,
+      "step": 26340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80311362286167e-05,
+      "loss": 1.1999,
+      "step": 26350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.803037457918869e-05,
+      "loss": 1.2797,
+      "step": 26360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8029612929760695e-05,
+      "loss": 1.1464,
+      "step": 26370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.802885128033269e-05,
+      "loss": 1.3164,
+      "step": 26380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802808963090469e-05,
+      "loss": 1.281,
+      "step": 26390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802732798147669e-05,
+      "loss": 1.2069,
+      "step": 26400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802656633204869e-05,
+      "loss": 1.2426,
+      "step": 26410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802580468262068e-05,
+      "loss": 1.2605,
+      "step": 26420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8025043033192686e-05,
+      "loss": 1.2609,
+      "step": 26430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8024281383764684e-05,
+      "loss": 1.1758,
+      "step": 26440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802351973433668e-05,
+      "loss": 1.1992,
+      "step": 26450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802275808490868e-05,
+      "loss": 1.1692,
+      "step": 26460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802199643548068e-05,
+      "loss": 1.3293,
+      "step": 26470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8021234786052674e-05,
+      "loss": 1.2629,
+      "step": 26480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.802047313662468e-05,
+      "loss": 1.3065,
+      "step": 26490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8019711487196675e-05,
+      "loss": 1.0884,
+      "step": 26500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801894983776867e-05,
+      "loss": 1.3394,
+      "step": 26510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801818818834068e-05,
+      "loss": 1.2494,
+      "step": 26520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8017426538912675e-05,
+      "loss": 1.2256,
+      "step": 26530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801666488948467e-05,
+      "loss": 1.0858,
+      "step": 26540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801590324005667e-05,
+      "loss": 1.1489,
+      "step": 26550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801514159062867e-05,
+      "loss": 1.2036,
+      "step": 26560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8014379941200664e-05,
+      "loss": 1.1879,
+      "step": 26570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801361829177267e-05,
+      "loss": 1.2452,
+      "step": 26580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.801285664234466e-05,
+      "loss": 1.4085,
+      "step": 26590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8012094992916664e-05,
+      "loss": 1.1832,
+      "step": 26600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.801133334348866e-05,
+      "loss": 1.1983,
+      "step": 26610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.801057169406066e-05,
+      "loss": 1.1996,
+      "step": 26620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.8009810044632656e-05,
+      "loss": 1.1481,
+      "step": 26630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800904839520466e-05,
+      "loss": 1.1695,
+      "step": 26640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800828674577665e-05,
+      "loss": 1.2557,
+      "step": 26650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.8007525096348655e-05,
+      "loss": 1.1998,
+      "step": 26660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800676344692065e-05,
+      "loss": 1.1929,
+      "step": 26670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800600179749265e-05,
+      "loss": 1.3724,
+      "step": 26680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800524014806465e-05,
+      "loss": 1.2197,
+      "step": 26690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800447849863665e-05,
+      "loss": 1.0455,
+      "step": 26700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800371684920865e-05,
+      "loss": 1.3798,
+      "step": 26710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800295519978065e-05,
+      "loss": 1.2479,
+      "step": 26720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800219355035265e-05,
+      "loss": 1.2366,
+      "step": 26730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.800143190092464e-05,
+      "loss": 1.0504,
+      "step": 26740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.8000670251496646e-05,
+      "loss": 1.2867,
+      "step": 26750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.799990860206864e-05,
+      "loss": 1.174,
+      "step": 26760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.799914695264064e-05,
+      "loss": 1.3134,
+      "step": 26770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.799838530321264e-05,
+      "loss": 1.1285,
+      "step": 26780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.799762365378464e-05,
+      "loss": 1.2104,
+      "step": 26790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.799686200435663e-05,
+      "loss": 1.3322,
+      "step": 26800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.799610035492864e-05,
+      "loss": 1.2212,
+      "step": 26810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7995338705500635e-05,
+      "loss": 1.1544,
+      "step": 26820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.799457705607263e-05,
+      "loss": 1.2336,
+      "step": 26830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.799381540664463e-05,
+      "loss": 1.1975,
+      "step": 26840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7993053757216634e-05,
+      "loss": 1.307,
+      "step": 26850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7992292107788625e-05,
+      "loss": 1.1607,
+      "step": 26860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.799153045836063e-05,
+      "loss": 1.1646,
+      "step": 26870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7990768808932627e-05,
+      "loss": 1.2365,
+      "step": 26880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7990007159504624e-05,
+      "loss": 1.2935,
+      "step": 26890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798924551007662e-05,
+      "loss": 1.2746,
+      "step": 26900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7988483860648626e-05,
+      "loss": 1.1025,
+      "step": 26910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798772221122062e-05,
+      "loss": 1.2281,
+      "step": 26920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798696056179262e-05,
+      "loss": 1.3795,
+      "step": 26930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7986198912364625e-05,
+      "loss": 1.183,
+      "step": 26940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7985437262936616e-05,
+      "loss": 1.1898,
+      "step": 26950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798467561350862e-05,
+      "loss": 1.2152,
+      "step": 26960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798391396408062e-05,
+      "loss": 1.1141,
+      "step": 26970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7983152314652615e-05,
+      "loss": 1.0914,
+      "step": 26980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798239066522461e-05,
+      "loss": 1.1996,
+      "step": 26990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7981629015796616e-05,
+      "loss": 1.2463,
+      "step": 27000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798086736636861e-05,
+      "loss": 1.3048,
+      "step": 27010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.798010571694061e-05,
+      "loss": 1.2273,
+      "step": 27020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.79793440675126e-05,
+      "loss": 1.1767,
+      "step": 27030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7978582418084606e-05,
+      "loss": 1.1356,
+      "step": 27040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7977820768656604e-05,
+      "loss": 1.2308,
+      "step": 27050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79770591192286e-05,
+      "loss": 1.2828,
+      "step": 27060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79762974698006e-05,
+      "loss": 1.1905,
+      "step": 27070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79755358203726e-05,
+      "loss": 1.0729,
+      "step": 27080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79747741709446e-05,
+      "loss": 1.1642,
+      "step": 27090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79740125215166e-05,
+      "loss": 1.2202,
+      "step": 27100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7973250872088595e-05,
+      "loss": 1.1321,
+      "step": 27110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.797248922266059e-05,
+      "loss": 1.3088,
+      "step": 27120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79717275732326e-05,
+      "loss": 1.1159,
+      "step": 27130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7970965923804595e-05,
+      "loss": 1.1067,
+      "step": 27140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.797020427437659e-05,
+      "loss": 1.159,
+      "step": 27150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.796944262494859e-05,
+      "loss": 1.3568,
+      "step": 27160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7968680975520594e-05,
+      "loss": 1.1588,
+      "step": 27170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7967919326092584e-05,
+      "loss": 1.1864,
+      "step": 27180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.796715767666459e-05,
+      "loss": 1.2408,
+      "step": 27190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7966396027236586e-05,
+      "loss": 1.2149,
+      "step": 27200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7965634377808584e-05,
+      "loss": 1.2605,
+      "step": 27210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.796487272838058e-05,
+      "loss": 1.1405,
+      "step": 27220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7964111078952585e-05,
+      "loss": 1.0989,
+      "step": 27230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7963349429524576e-05,
+      "loss": 1.3337,
+      "step": 27240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.796258778009658e-05,
+      "loss": 1.2312,
+      "step": 27250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.796182613066858e-05,
+      "loss": 1.1958,
+      "step": 27260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7961064481240575e-05,
+      "loss": 1.2614,
+      "step": 27270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.796030283181257e-05,
+      "loss": 1.1523,
+      "step": 27280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795954118238458e-05,
+      "loss": 1.2033,
+      "step": 27290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7958779532956574e-05,
+      "loss": 1.0476,
+      "step": 27300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795801788352857e-05,
+      "loss": 1.1526,
+      "step": 27310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7957256234100576e-05,
+      "loss": 1.3308,
+      "step": 27320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795649458467257e-05,
+      "loss": 1.2664,
+      "step": 27330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795573293524457e-05,
+      "loss": 1.2091,
+      "step": 27340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795497128581657e-05,
+      "loss": 1.2311,
+      "step": 27350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7954209636388566e-05,
+      "loss": 1.1484,
+      "step": 27360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795344798696056e-05,
+      "loss": 1.0877,
+      "step": 27370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795268633753257e-05,
+      "loss": 1.2063,
+      "step": 27380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795192468810456e-05,
+      "loss": 1.17,
+      "step": 27390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795116303867656e-05,
+      "loss": 1.1934,
+      "step": 27400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.795040138924856e-05,
+      "loss": 1.2019,
+      "step": 27410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.794963973982056e-05,
+      "loss": 1.2161,
+      "step": 27420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7948878090392555e-05,
+      "loss": 1.191,
+      "step": 27430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.794811644096456e-05,
+      "loss": 1.2565,
+      "step": 27440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.794735479153655e-05,
+      "loss": 1.3532,
+      "step": 27450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7946593142108554e-05,
+      "loss": 1.1115,
+      "step": 27460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.794583149268055e-05,
+      "loss": 1.2019,
+      "step": 27470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.794506984325255e-05,
+      "loss": 1.0874,
+      "step": 27480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7944308193824547e-05,
+      "loss": 1.3029,
+      "step": 27490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7943546544396544e-05,
+      "loss": 1.3988,
+      "step": 27500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.794278489496855e-05,
+      "loss": 1.1652,
+      "step": 27510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7942023245540546e-05,
+      "loss": 1.3175,
+      "step": 27520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.794126159611254e-05,
+      "loss": 1.2016,
+      "step": 27530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.794049994668454e-05,
+      "loss": 1.0668,
+      "step": 27540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7939738297256545e-05,
+      "loss": 1.1303,
+      "step": 27550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7938976647828536e-05,
+      "loss": 1.1379,
+      "step": 27560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793821499840054e-05,
+      "loss": 1.0776,
+      "step": 27570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793745334897254e-05,
+      "loss": 1.3399,
+      "step": 27580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7936691699544535e-05,
+      "loss": 1.238,
+      "step": 27590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793593005011653e-05,
+      "loss": 1.2388,
+      "step": 27600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7935168400688536e-05,
+      "loss": 1.2603,
+      "step": 27610
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793440675126053e-05,
+      "loss": 1.0882,
+      "step": 27620
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793364510183253e-05,
+      "loss": 1.1854,
+      "step": 27630
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793288345240453e-05,
+      "loss": 1.0878,
+      "step": 27640
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7932121802976526e-05,
+      "loss": 1.2738,
+      "step": 27650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7931360153548524e-05,
+      "loss": 1.2672,
+      "step": 27660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.793059850412053e-05,
+      "loss": 1.4292,
+      "step": 27670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7929836854692525e-05,
+      "loss": 1.2026,
+      "step": 27680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.792907520526452e-05,
+      "loss": 1.1884,
+      "step": 27690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.792831355583652e-05,
+      "loss": 1.1173,
+      "step": 27700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792755190640852e-05,
+      "loss": 1.2783,
+      "step": 27710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792679025698052e-05,
+      "loss": 1.1209,
+      "step": 27720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792602860755252e-05,
+      "loss": 1.25,
+      "step": 27730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792526695812452e-05,
+      "loss": 1.2723,
+      "step": 27740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7924505308696515e-05,
+      "loss": 1.0484,
+      "step": 27750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792374365926852e-05,
+      "loss": 1.2193,
+      "step": 27760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792298200984051e-05,
+      "loss": 1.1915,
+      "step": 27770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7922220360412514e-05,
+      "loss": 1.0945,
+      "step": 27780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792145871098451e-05,
+      "loss": 1.2852,
+      "step": 27790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.792069706155651e-05,
+      "loss": 1.0895,
+      "step": 27800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7919935412128506e-05,
+      "loss": 1.1353,
+      "step": 27810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.791917376270051e-05,
+      "loss": 1.1268,
+      "step": 27820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79184121132725e-05,
+      "loss": 1.109,
+      "step": 27830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7917650463844505e-05,
+      "loss": 1.2813,
+      "step": 27840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79168888144165e-05,
+      "loss": 1.0984,
+      "step": 27850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79161271649885e-05,
+      "loss": 1.1169,
+      "step": 27860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79153655155605e-05,
+      "loss": 1.192,
+      "step": 27870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79146038661325e-05,
+      "loss": 1.2346,
+      "step": 27880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79138422167045e-05,
+      "loss": 1.0941,
+      "step": 27890
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.79130805672765e-05,
+      "loss": 1.3112,
+      "step": 27900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7912318917848494e-05,
+      "loss": 1.1944,
+      "step": 27910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.791155726842049e-05,
+      "loss": 1.1721,
+      "step": 27920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7910795618992496e-05,
+      "loss": 1.188,
+      "step": 27930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.791003396956449e-05,
+      "loss": 1.2407,
+      "step": 27940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790927232013649e-05,
+      "loss": 1.1182,
+      "step": 27950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790851067070849e-05,
+      "loss": 1.1782,
+      "step": 27960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7907749021280486e-05,
+      "loss": 0.9735,
+      "step": 27970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790698737185248e-05,
+      "loss": 1.1812,
+      "step": 27980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790622572242449e-05,
+      "loss": 1.2219,
+      "step": 27990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790546407299648e-05,
+      "loss": 1.2381,
+      "step": 28000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790470242356848e-05,
+      "loss": 1.2806,
+      "step": 28010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790394077414048e-05,
+      "loss": 1.1199,
+      "step": 28020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790317912471248e-05,
+      "loss": 1.0935,
+      "step": 28030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7902417475284475e-05,
+      "loss": 1.1268,
+      "step": 28040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790165582585648e-05,
+      "loss": 1.077,
+      "step": 28050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.790089417642848e-05,
+      "loss": 1.2266,
+      "step": 28060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7900132527000474e-05,
+      "loss": 1.2037,
+      "step": 28070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.789937087757247e-05,
+      "loss": 1.1895,
+      "step": 28080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.789860922814447e-05,
+      "loss": 1.1503,
+      "step": 28090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.789784757871647e-05,
+      "loss": 1.0915,
+      "step": 28100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.789708592928847e-05,
+      "loss": 1.204,
+      "step": 28110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.789632427986047e-05,
+      "loss": 1.1382,
+      "step": 28120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7895562630432466e-05,
+      "loss": 1.1423,
+      "step": 28130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.789480098100447e-05,
+      "loss": 1.269,
+      "step": 28140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.789403933157646e-05,
+      "loss": 1.2586,
+      "step": 28150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7893277682148465e-05,
+      "loss": 1.1661,
+      "step": 28160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.789251603272046e-05,
+      "loss": 1.3663,
+      "step": 28170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.789175438329246e-05,
+      "loss": 1.4328,
+      "step": 28180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.789099273386446e-05,
+      "loss": 1.2208,
+      "step": 28190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.789023108443646e-05,
+      "loss": 1.2014,
+      "step": 28200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788946943500845e-05,
+      "loss": 1.276,
+      "step": 28210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7888707785580456e-05,
+      "loss": 1.2458,
+      "step": 28220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7887946136152454e-05,
+      "loss": 1.2453,
+      "step": 28230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788718448672445e-05,
+      "loss": 1.0857,
+      "step": 28240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788642283729645e-05,
+      "loss": 1.0752,
+      "step": 28250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788566118786845e-05,
+      "loss": 1.1631,
+      "step": 28260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788489953844045e-05,
+      "loss": 1.1017,
+      "step": 28270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788413788901245e-05,
+      "loss": 1.3083,
+      "step": 28280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7883376239584445e-05,
+      "loss": 1.1876,
+      "step": 28290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788261459015644e-05,
+      "loss": 1.1663,
+      "step": 28300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788185294072845e-05,
+      "loss": 1.3264,
+      "step": 28310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7881091291300445e-05,
+      "loss": 1.1719,
+      "step": 28320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.788032964187244e-05,
+      "loss": 1.1144,
+      "step": 28330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.787956799244444e-05,
+      "loss": 1.1885,
+      "step": 28340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7878806343016444e-05,
+      "loss": 1.2677,
+      "step": 28350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7878044693588435e-05,
+      "loss": 1.2536,
+      "step": 28360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.787728304416044e-05,
+      "loss": 1.3236,
+      "step": 28370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7876521394732436e-05,
+      "loss": 1.2753,
+      "step": 28380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7875759745304434e-05,
+      "loss": 1.1973,
+      "step": 28390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.787499809587643e-05,
+      "loss": 1.21,
+      "step": 28400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.787423644644843e-05,
+      "loss": 1.2143,
+      "step": 28410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7873474797020426e-05,
+      "loss": 1.2209,
+      "step": 28420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.787271314759243e-05,
+      "loss": 1.1633,
+      "step": 28430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.787195149816443e-05,
+      "loss": 1.1898,
+      "step": 28440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7871189848736425e-05,
+      "loss": 1.1521,
+      "step": 28450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.787042819930842e-05,
+      "loss": 1.1984,
+      "step": 28460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786966654988042e-05,
+      "loss": 1.1295,
+      "step": 28470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7868904900452424e-05,
+      "loss": 1.146,
+      "step": 28480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786814325102442e-05,
+      "loss": 1.1341,
+      "step": 28490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786738160159642e-05,
+      "loss": 1.2759,
+      "step": 28500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786661995216842e-05,
+      "loss": 1.0714,
+      "step": 28510
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786585830274042e-05,
+      "loss": 1.0784,
+      "step": 28520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786509665331241e-05,
+      "loss": 1.1527,
+      "step": 28530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7864335003884416e-05,
+      "loss": 1.1806,
+      "step": 28540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7863573354456414e-05,
+      "loss": 1.1263,
+      "step": 28550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786281170502841e-05,
+      "loss": 1.2754,
+      "step": 28560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.786205005560041e-05,
+      "loss": 1.2284,
+      "step": 28570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.786128840617241e-05,
+      "loss": 1.1944,
+      "step": 28580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78605267567444e-05,
+      "loss": 1.2795,
+      "step": 28590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.785976510731641e-05,
+      "loss": 1.135,
+      "step": 28600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7859003457888405e-05,
+      "loss": 1.0458,
+      "step": 28610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78582418084604e-05,
+      "loss": 1.2073,
+      "step": 28620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78574801590324e-05,
+      "loss": 1.2368,
+      "step": 28630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7856718509604404e-05,
+      "loss": 1.2034,
+      "step": 28640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78559568601764e-05,
+      "loss": 1.1603,
+      "step": 28650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78551952107484e-05,
+      "loss": 0.9389,
+      "step": 28660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78544335613204e-05,
+      "loss": 1.1833,
+      "step": 28670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7853671911892394e-05,
+      "loss": 1.0315,
+      "step": 28680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78529102624644e-05,
+      "loss": 1.2857,
+      "step": 28690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7852148613036396e-05,
+      "loss": 1.1944,
+      "step": 28700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.785138696360839e-05,
+      "loss": 1.1109,
+      "step": 28710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.785062531418039e-05,
+      "loss": 1.0567,
+      "step": 28720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7849863664752395e-05,
+      "loss": 1.1965,
+      "step": 28730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7849102015324386e-05,
+      "loss": 1.2013,
+      "step": 28740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.784834036589639e-05,
+      "loss": 1.0881,
+      "step": 28750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.784757871646839e-05,
+      "loss": 1.2287,
+      "step": 28760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7846817067040385e-05,
+      "loss": 1.226,
+      "step": 28770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.784605541761238e-05,
+      "loss": 1.2432,
+      "step": 28780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7845293768184387e-05,
+      "loss": 1.2143,
+      "step": 28790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.784453211875638e-05,
+      "loss": 1.1126,
+      "step": 28800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.784377046932838e-05,
+      "loss": 1.2087,
+      "step": 28810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.784300881990038e-05,
+      "loss": 1.1965,
+      "step": 28820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7842247170472376e-05,
+      "loss": 1.1745,
+      "step": 28830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7841485521044374e-05,
+      "loss": 1.2149,
+      "step": 28840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.784072387161638e-05,
+      "loss": 1.2292,
+      "step": 28850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7839962222188376e-05,
+      "loss": 1.1644,
+      "step": 28860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783920057276037e-05,
+      "loss": 1.29,
+      "step": 28870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783843892333237e-05,
+      "loss": 1.1545,
+      "step": 28880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783767727390437e-05,
+      "loss": 1.2241,
+      "step": 28890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783691562447637e-05,
+      "loss": 1.2698,
+      "step": 28900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783615397504836e-05,
+      "loss": 1.3073,
+      "step": 28910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783539232562037e-05,
+      "loss": 1.163,
+      "step": 28920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7834630676192365e-05,
+      "loss": 1.2401,
+      "step": 28930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783386902676436e-05,
+      "loss": 1.1244,
+      "step": 28940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783310737733636e-05,
+      "loss": 1.2213,
+      "step": 28950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7832345727908364e-05,
+      "loss": 1.1851,
+      "step": 28960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7831584078480355e-05,
+      "loss": 0.9937,
+      "step": 28970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.783082242905236e-05,
+      "loss": 1.1157,
+      "step": 28980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7830060779624356e-05,
+      "loss": 1.2044,
+      "step": 28990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7829299130196354e-05,
+      "loss": 1.4413,
+      "step": 29000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.782853748076835e-05,
+      "loss": 1.2262,
+      "step": 29010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7827775831340355e-05,
+      "loss": 1.1887,
+      "step": 29020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782701418191235e-05,
+      "loss": 1.2611,
+      "step": 29030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782625253248435e-05,
+      "loss": 1.0608,
+      "step": 29040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782549088305635e-05,
+      "loss": 1.0451,
+      "step": 29050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7824729233628345e-05,
+      "loss": 1.165,
+      "step": 29060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782396758420035e-05,
+      "loss": 1.2086,
+      "step": 29070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782320593477235e-05,
+      "loss": 1.1588,
+      "step": 29080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7822444285344344e-05,
+      "loss": 1.1244,
+      "step": 29090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782168263591634e-05,
+      "loss": 1.171,
+      "step": 29100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7820920986488346e-05,
+      "loss": 1.1462,
+      "step": 29110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.782015933706034e-05,
+      "loss": 1.2239,
+      "step": 29120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781939768763234e-05,
+      "loss": 1.2208,
+      "step": 29130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781863603820434e-05,
+      "loss": 1.1141,
+      "step": 29140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7817874388776336e-05,
+      "loss": 1.2294,
+      "step": 29150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7817112739348334e-05,
+      "loss": 1.1028,
+      "step": 29160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781635108992034e-05,
+      "loss": 1.2139,
+      "step": 29170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781558944049233e-05,
+      "loss": 1.2677,
+      "step": 29180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781482779106433e-05,
+      "loss": 1.1713,
+      "step": 29190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781406614163633e-05,
+      "loss": 1.1102,
+      "step": 29200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781330449220833e-05,
+      "loss": 1.189,
+      "step": 29210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7812542842780325e-05,
+      "loss": 1.2051,
+      "step": 29220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.781178119335233e-05,
+      "loss": 1.1583,
+      "step": 29230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.781101954392433e-05,
+      "loss": 1.1589,
+      "step": 29240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7810257894496324e-05,
+      "loss": 1.2277,
+      "step": 29250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780949624506832e-05,
+      "loss": 1.2491,
+      "step": 29260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780873459564032e-05,
+      "loss": 1.1454,
+      "step": 29270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7807972946212323e-05,
+      "loss": 1.2339,
+      "step": 29280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780721129678432e-05,
+      "loss": 1.1212,
+      "step": 29290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780644964735632e-05,
+      "loss": 1.0594,
+      "step": 29300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7805687997928316e-05,
+      "loss": 1.234,
+      "step": 29310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780492634850031e-05,
+      "loss": 1.2005,
+      "step": 29320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780416469907231e-05,
+      "loss": 1.2192,
+      "step": 29330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7803403049644315e-05,
+      "loss": 1.213,
+      "step": 29340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7802641400216306e-05,
+      "loss": 1.3816,
+      "step": 29350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780187975078831e-05,
+      "loss": 1.1083,
+      "step": 29360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.780111810136031e-05,
+      "loss": 1.2151,
+      "step": 29370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7800356451932305e-05,
+      "loss": 1.2318,
+      "step": 29380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77995948025043e-05,
+      "loss": 1.2243,
+      "step": 29390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7798833153076307e-05,
+      "loss": 1.2702,
+      "step": 29400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77980715036483e-05,
+      "loss": 1.1399,
+      "step": 29410
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77973098542203e-05,
+      "loss": 1.1926,
+      "step": 29420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77965482047923e-05,
+      "loss": 1.0885,
+      "step": 29430
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7795786555364296e-05,
+      "loss": 1.2326,
+      "step": 29440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77950249059363e-05,
+      "loss": 1.2278,
+      "step": 29450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.77942632565083e-05,
+      "loss": 1.1037,
+      "step": 29460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7793501607080296e-05,
+      "loss": 1.1793,
+      "step": 29470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.779273995765229e-05,
+      "loss": 1.1204,
+      "step": 29480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.77919783082243e-05,
+      "loss": 1.2366,
+      "step": 29490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.779121665879629e-05,
+      "loss": 1.0798,
+      "step": 29500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.779045500936829e-05,
+      "loss": 1.1859,
+      "step": 29510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778969335994029e-05,
+      "loss": 1.126,
+      "step": 29520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778893171051229e-05,
+      "loss": 1.2093,
+      "step": 29530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7788170061084285e-05,
+      "loss": 1.2219,
+      "step": 29540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778740841165629e-05,
+      "loss": 1.1287,
+      "step": 29550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778664676222828e-05,
+      "loss": 1.2222,
+      "step": 29560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7785885112800284e-05,
+      "loss": 1.1762,
+      "step": 29570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778512346337228e-05,
+      "loss": 1.1822,
+      "step": 29580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778436181394428e-05,
+      "loss": 1.0881,
+      "step": 29590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7783600164516276e-05,
+      "loss": 1.1409,
+      "step": 29600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778283851508828e-05,
+      "loss": 1.388,
+      "step": 29610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778207686566028e-05,
+      "loss": 1.1889,
+      "step": 29620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7781315216232275e-05,
+      "loss": 1.1552,
+      "step": 29630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.778055356680427e-05,
+      "loss": 1.2775,
+      "step": 29640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.777979191737627e-05,
+      "loss": 1.1934,
+      "step": 29650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7779030267948275e-05,
+      "loss": 1.2758,
+      "step": 29660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.777826861852027e-05,
+      "loss": 1.1207,
+      "step": 29670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777750696909227e-05,
+      "loss": 1.3277,
+      "step": 29680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777674531966427e-05,
+      "loss": 1.1202,
+      "step": 29690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777598367023627e-05,
+      "loss": 1.0646,
+      "step": 29700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777522202080826e-05,
+      "loss": 1.2702,
+      "step": 29710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7774460371380266e-05,
+      "loss": 1.2543,
+      "step": 29720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7773698721952264e-05,
+      "loss": 1.1946,
+      "step": 29730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777293707252426e-05,
+      "loss": 1.1307,
+      "step": 29740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777217542309626e-05,
+      "loss": 1.2123,
+      "step": 29750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.777141377366826e-05,
+      "loss": 1.2229,
+      "step": 29760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7770652124240254e-05,
+      "loss": 1.1789,
+      "step": 29770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776989047481226e-05,
+      "loss": 1.4337,
+      "step": 29780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776912882538425e-05,
+      "loss": 1.2461,
+      "step": 29790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776836717595625e-05,
+      "loss": 1.2883,
+      "step": 29800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776760552652825e-05,
+      "loss": 1.0738,
+      "step": 29810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776684387710025e-05,
+      "loss": 1.2898,
+      "step": 29820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776608222767225e-05,
+      "loss": 1.0916,
+      "step": 29830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776532057824425e-05,
+      "loss": 1.2577,
+      "step": 29840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776455892881625e-05,
+      "loss": 1.0704,
+      "step": 29850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7763797279388244e-05,
+      "loss": 1.2133,
+      "step": 29860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776303562996025e-05,
+      "loss": 1.2434,
+      "step": 29870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.776227398053224e-05,
+      "loss": 1.108,
+      "step": 29880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7761512331104243e-05,
+      "loss": 1.2859,
+      "step": 29890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.776075068167624e-05,
+      "loss": 1.247,
+      "step": 29900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775998903224824e-05,
+      "loss": 1.2783,
+      "step": 29910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7759227382820236e-05,
+      "loss": 1.0678,
+      "step": 29920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775846573339224e-05,
+      "loss": 1.1202,
+      "step": 29930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775770408396423e-05,
+      "loss": 1.1974,
+      "step": 29940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7756942434536235e-05,
+      "loss": 1.1407,
+      "step": 29950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775618078510823e-05,
+      "loss": 1.187,
+      "step": 29960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775541913568023e-05,
+      "loss": 1.2128,
+      "step": 29970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775465748625223e-05,
+      "loss": 1.3105,
+      "step": 29980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775389583682423e-05,
+      "loss": 1.2086,
+      "step": 29990
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775313418739622e-05,
+      "loss": 1.2636,
+      "step": 30000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775237253796823e-05,
+      "loss": 1.1848,
+      "step": 30010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7751610888540224e-05,
+      "loss": 1.2386,
+      "step": 30020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775084923911222e-05,
+      "loss": 1.2038,
+      "step": 30030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7750087589684226e-05,
+      "loss": 1.2592,
+      "step": 30040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774932594025622e-05,
+      "loss": 1.1675,
+      "step": 30050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774856429082822e-05,
+      "loss": 1.2018,
+      "step": 30060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774780264140022e-05,
+      "loss": 1.2285,
+      "step": 30070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774704099197222e-05,
+      "loss": 1.303,
+      "step": 30080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774627934254421e-05,
+      "loss": 1.3199,
+      "step": 30090
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774551769311622e-05,
+      "loss": 1.1402,
+      "step": 30100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7744756043688215e-05,
+      "loss": 1.2757,
+      "step": 30110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.774399439426021e-05,
+      "loss": 1.1746,
+      "step": 30120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.774323274483221e-05,
+      "loss": 1.3049,
+      "step": 30130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7742471095404214e-05,
+      "loss": 1.1911,
+      "step": 30140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7741709445976205e-05,
+      "loss": 1.193,
+      "step": 30150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.774094779654821e-05,
+      "loss": 1.1257,
+      "step": 30160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7740186147120206e-05,
+      "loss": 1.3176,
+      "step": 30170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7739424497692204e-05,
+      "loss": 1.1976,
+      "step": 30180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.77386628482642e-05,
+      "loss": 1.3913,
+      "step": 30190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7737901198836206e-05,
+      "loss": 1.2573,
+      "step": 30200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7737139549408196e-05,
+      "loss": 1.1921,
+      "step": 30210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.77363778999802e-05,
+      "loss": 1.2742,
+      "step": 30220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.77356162505522e-05,
+      "loss": 1.3111,
+      "step": 30230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7734854601124195e-05,
+      "loss": 1.201,
+      "step": 30240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.77340929516962e-05,
+      "loss": 1.1184,
+      "step": 30250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.773333130226819e-05,
+      "loss": 1.1837,
+      "step": 30260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7732569652840195e-05,
+      "loss": 1.1781,
+      "step": 30270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.773180800341219e-05,
+      "loss": 1.1919,
+      "step": 30280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.773104635398419e-05,
+      "loss": 1.0355,
+      "step": 30290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.773028470455619e-05,
+      "loss": 1.0351,
+      "step": 30300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.772952305512819e-05,
+      "loss": 1.1007,
+      "step": 30310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.772876140570018e-05,
+      "loss": 1.2022,
+      "step": 30320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7727999756272186e-05,
+      "loss": 1.2913,
+      "step": 30330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7727238106844184e-05,
+      "loss": 1.2997,
+      "step": 30340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.772647645741618e-05,
+      "loss": 1.1148,
+      "step": 30350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.772571480798818e-05,
+      "loss": 1.1129,
+      "step": 30360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.772495315856018e-05,
+      "loss": 1.1537,
+      "step": 30370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7724191509132174e-05,
+      "loss": 1.2211,
+      "step": 30380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.772342985970418e-05,
+      "loss": 1.1982,
+      "step": 30390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7722668210276175e-05,
+      "loss": 1.2122,
+      "step": 30400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.772190656084817e-05,
+      "loss": 1.2977,
+      "step": 30410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.772114491142018e-05,
+      "loss": 1.1895,
+      "step": 30420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7720383261992174e-05,
+      "loss": 1.2022,
+      "step": 30430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.771962161256417e-05,
+      "loss": 1.2391,
+      "step": 30440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.771885996313617e-05,
+      "loss": 1.1713,
+      "step": 30450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7718098313708174e-05,
+      "loss": 1.0567,
+      "step": 30460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7717336664280164e-05,
+      "loss": 1.3388,
+      "step": 30470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.771657501485217e-05,
+      "loss": 1.1904,
+      "step": 30480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7715813365424166e-05,
+      "loss": 1.0896,
+      "step": 30490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7715051715996163e-05,
+      "loss": 1.2439,
+      "step": 30500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.771429006656816e-05,
+      "loss": 1.1697,
+      "step": 30510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7713528417140165e-05,
+      "loss": 1.1787,
+      "step": 30520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7712766767712156e-05,
+      "loss": 1.0922,
+      "step": 30530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.771200511828416e-05,
+      "loss": 1.2386,
+      "step": 30540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.771124346885616e-05,
+      "loss": 1.1014,
+      "step": 30550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7710481819428155e-05,
+      "loss": 1.165,
+      "step": 30560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770972017000015e-05,
+      "loss": 1.0874,
+      "step": 30570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770895852057216e-05,
+      "loss": 1.1702,
+      "step": 30580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770819687114415e-05,
+      "loss": 1.2388,
+      "step": 30590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770743522171615e-05,
+      "loss": 1.1178,
+      "step": 30600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770667357228815e-05,
+      "loss": 1.0906,
+      "step": 30610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770591192286015e-05,
+      "loss": 1.1356,
+      "step": 30620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770515027343215e-05,
+      "loss": 1.2354,
+      "step": 30630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770438862400415e-05,
+      "loss": 1.227,
+      "step": 30640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7703626974576146e-05,
+      "loss": 1.2091,
+      "step": 30650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770286532514814e-05,
+      "loss": 1.0696,
+      "step": 30660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770210367572015e-05,
+      "loss": 1.1081,
+      "step": 30670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770134202629214e-05,
+      "loss": 1.1629,
+      "step": 30680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.770058037686414e-05,
+      "loss": 1.232,
+      "step": 30690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.769981872743613e-05,
+      "loss": 1.1243,
+      "step": 30700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.769905707800814e-05,
+      "loss": 1.148,
+      "step": 30710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7698295428580135e-05,
+      "loss": 1.1105,
+      "step": 30720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.769753377915213e-05,
+      "loss": 1.3192,
+      "step": 30730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.769677212972413e-05,
+      "loss": 1.1628,
+      "step": 30740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7696010480296134e-05,
+      "loss": 1.179,
+      "step": 30750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7695248830868125e-05,
+      "loss": 1.1129,
+      "step": 30760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.769448718144013e-05,
+      "loss": 1.0899,
+      "step": 30770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7693725532012126e-05,
+      "loss": 1.1229,
+      "step": 30780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7692963882584124e-05,
+      "loss": 1.0426,
+      "step": 30790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.769220223315612e-05,
+      "loss": 1.1017,
+      "step": 30800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7691440583728126e-05,
+      "loss": 1.3757,
+      "step": 30810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.769067893430012e-05,
+      "loss": 1.065,
+      "step": 30820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768991728487212e-05,
+      "loss": 1.2196,
+      "step": 30830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7689155635444125e-05,
+      "loss": 1.09,
+      "step": 30840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7688393986016115e-05,
+      "loss": 1.1728,
+      "step": 30850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768763233658812e-05,
+      "loss": 1.2281,
+      "step": 30860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768687068716012e-05,
+      "loss": 1.2728,
+      "step": 30870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7686109037732115e-05,
+      "loss": 1.2716,
+      "step": 30880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768534738830411e-05,
+      "loss": 1.0865,
+      "step": 30890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7684585738876116e-05,
+      "loss": 1.153,
+      "step": 30900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768382408944811e-05,
+      "loss": 1.303,
+      "step": 30910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768306244002011e-05,
+      "loss": 1.2467,
+      "step": 30920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768230079059211e-05,
+      "loss": 1.1629,
+      "step": 30930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7681539141164106e-05,
+      "loss": 1.1279,
+      "step": 30940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7680777491736104e-05,
+      "loss": 1.26,
+      "step": 30950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.768001584230811e-05,
+      "loss": 1.1015,
+      "step": 30960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.76792541928801e-05,
+      "loss": 1.2798,
+      "step": 30970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.76784925434521e-05,
+      "loss": 1.198,
+      "step": 30980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.76777308940241e-05,
+      "loss": 1.0794,
+      "step": 30990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.76769692445961e-05,
+      "loss": 1.1905,
+      "step": 31000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7676207595168095e-05,
+      "loss": 1.1407,
+      "step": 31010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.76754459457401e-05,
+      "loss": 1.1256,
+      "step": 31020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.76746842963121e-05,
+      "loss": 1.2068,
+      "step": 31030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7673922646884094e-05,
+      "loss": 1.1882,
+      "step": 31040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.76731609974561e-05,
+      "loss": 1.2311,
+      "step": 31050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.767239934802809e-05,
+      "loss": 1.1232,
+      "step": 31060
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7671637698600094e-05,
+      "loss": 1.2395,
+      "step": 31070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.767087604917209e-05,
+      "loss": 1.0229,
+      "step": 31080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.767011439974409e-05,
+      "loss": 1.2699,
+      "step": 31090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7669352750316086e-05,
+      "loss": 1.1505,
+      "step": 31100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766859110088809e-05,
+      "loss": 1.2853,
+      "step": 31110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766782945146008e-05,
+      "loss": 1.1511,
+      "step": 31120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7667067802032085e-05,
+      "loss": 1.2135,
+      "step": 31130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766630615260408e-05,
+      "loss": 1.022,
+      "step": 31140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766554450317608e-05,
+      "loss": 1.0955,
+      "step": 31150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766478285374808e-05,
+      "loss": 1.2615,
+      "step": 31160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7664021204320075e-05,
+      "loss": 1.3165,
+      "step": 31170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766325955489207e-05,
+      "loss": 1.2059,
+      "step": 31180
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.766249790546408e-05,
+      "loss": 1.1236,
+      "step": 31190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7661736256036074e-05,
+      "loss": 1.2386,
+      "step": 31200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.766097460660807e-05,
+      "loss": 1.2992,
+      "step": 31210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7660212957180076e-05,
+      "loss": 1.1523,
+      "step": 31220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765945130775207e-05,
+      "loss": 1.1228,
+      "step": 31230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765868965832407e-05,
+      "loss": 1.0413,
+      "step": 31240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765792800889607e-05,
+      "loss": 1.3144,
+      "step": 31250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7657166359468066e-05,
+      "loss": 1.3165,
+      "step": 31260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765640471004006e-05,
+      "loss": 1.2142,
+      "step": 31270
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765564306061207e-05,
+      "loss": 1.0695,
+      "step": 31280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765488141118406e-05,
+      "loss": 1.1281,
+      "step": 31290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765411976175606e-05,
+      "loss": 1.1606,
+      "step": 31300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765335811232806e-05,
+      "loss": 1.1413,
+      "step": 31310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765259646290006e-05,
+      "loss": 1.2275,
+      "step": 31320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7651834813472055e-05,
+      "loss": 1.2,
+      "step": 31330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765107316404406e-05,
+      "loss": 1.2259,
+      "step": 31340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.765031151461605e-05,
+      "loss": 1.2017,
+      "step": 31350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7649549865188054e-05,
+      "loss": 1.3814,
+      "step": 31360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.764878821576005e-05,
+      "loss": 1.2751,
+      "step": 31370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.764802656633205e-05,
+      "loss": 1.0704,
+      "step": 31380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7647264916904046e-05,
+      "loss": 1.1718,
+      "step": 31390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.764650326747605e-05,
+      "loss": 1.2842,
+      "step": 31400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.764574161804805e-05,
+      "loss": 1.3122,
+      "step": 31410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7644979968620046e-05,
+      "loss": 1.113,
+      "step": 31420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.764421831919205e-05,
+      "loss": 1.302,
+      "step": 31430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.764345666976404e-05,
+      "loss": 1.249,
+      "step": 31440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7642695020336045e-05,
+      "loss": 1.2712,
+      "step": 31450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.764193337090804e-05,
+      "loss": 1.1282,
+      "step": 31460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.764117172148004e-05,
+      "loss": 1.2476,
+      "step": 31470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.764041007205204e-05,
+      "loss": 1.1487,
+      "step": 31480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763964842262404e-05,
+      "loss": 1.2346,
+      "step": 31490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763888677319603e-05,
+      "loss": 1.2287,
+      "step": 31500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7638125123768036e-05,
+      "loss": 1.065,
+      "step": 31510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7637363474340034e-05,
+      "loss": 1.1028,
+      "step": 31520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763660182491203e-05,
+      "loss": 1.1346,
+      "step": 31530
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763584017548403e-05,
+      "loss": 1.1007,
+      "step": 31540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763507852605603e-05,
+      "loss": 1.2259,
+      "step": 31550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7634316876628024e-05,
+      "loss": 1.0159,
+      "step": 31560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763355522720003e-05,
+      "loss": 1.1643,
+      "step": 31570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7632793577772025e-05,
+      "loss": 1.0646,
+      "step": 31580
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763203192834402e-05,
+      "loss": 1.2153,
+      "step": 31590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.763127027891602e-05,
+      "loss": 1.1376,
+      "step": 31600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7630508629488025e-05,
+      "loss": 1.1398,
+      "step": 31610
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.762974698006002e-05,
+      "loss": 1.2004,
+      "step": 31620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.762898533063202e-05,
+      "loss": 1.079,
+      "step": 31630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.762822368120402e-05,
+      "loss": 1.1667,
+      "step": 31640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7627462031776014e-05,
+      "loss": 1.1491,
+      "step": 31650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.762670038234802e-05,
+      "loss": 1.2274,
+      "step": 31660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.762593873292001e-05,
+      "loss": 1.1001,
+      "step": 31670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7625177083492014e-05,
+      "loss": 1.0699,
+      "step": 31680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.762441543406401e-05,
+      "loss": 1.1859,
+      "step": 31690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.762365378463601e-05,
+      "loss": 1.2064,
+      "step": 31700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7622892135208006e-05,
+      "loss": 1.234,
+      "step": 31710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.762213048578001e-05,
+      "loss": 1.1896,
+      "step": 31720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7621368836352e-05,
+      "loss": 1.1885,
+      "step": 31730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7620607186924005e-05,
+      "loss": 1.2648,
+      "step": 31740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7619845537496e-05,
+      "loss": 1.2439,
+      "step": 31750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7619083888068e-05,
+      "loss": 1.4709,
+      "step": 31760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.761832223864e-05,
+      "loss": 1.3192,
+      "step": 31770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7617560589212e-05,
+      "loss": 1.1301,
+      "step": 31780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7616798939784e-05,
+      "loss": 1.1451,
+      "step": 31790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7616037290356e-05,
+      "loss": 1.2441,
+      "step": 31800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7615275640927994e-05,
+      "loss": 1.1093,
+      "step": 31810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.761451399149999e-05,
+      "loss": 1.2189,
+      "step": 31820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7613752342071996e-05,
+      "loss": 1.1305,
+      "step": 31830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7612990692643993e-05,
+      "loss": 1.2537,
+      "step": 31840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.761222904321599e-05,
+      "loss": 1.165,
+      "step": 31850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.761146739378799e-05,
+      "loss": 1.2528,
+      "step": 31860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.761070574435999e-05,
+      "loss": 1.1189,
+      "step": 31870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760994409493198e-05,
+      "loss": 1.2943,
+      "step": 31880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760918244550399e-05,
+      "loss": 1.1433,
+      "step": 31890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7608420796075985e-05,
+      "loss": 1.1831,
+      "step": 31900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760765914664798e-05,
+      "loss": 1.3833,
+      "step": 31910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760689749721998e-05,
+      "loss": 1.2153,
+      "step": 31920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7606135847791984e-05,
+      "loss": 1.2194,
+      "step": 31930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7605374198363975e-05,
+      "loss": 1.0869,
+      "step": 31940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760461254893598e-05,
+      "loss": 1.218,
+      "step": 31950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7603850899507977e-05,
+      "loss": 1.1739,
+      "step": 31960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7603089250079974e-05,
+      "loss": 1.1341,
+      "step": 31970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760232760065197e-05,
+      "loss": 1.1321,
+      "step": 31980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7601565951223976e-05,
+      "loss": 1.226,
+      "step": 31990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760080430179597e-05,
+      "loss": 1.0231,
+      "step": 32000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.760004265236797e-05,
+      "loss": 1.2377,
+      "step": 32010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7599281002939975e-05,
+      "loss": 1.0834,
+      "step": 32020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7598519353511966e-05,
+      "loss": 1.4138,
+      "step": 32030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.759775770408397e-05,
+      "loss": 1.2797,
+      "step": 32040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.759699605465597e-05,
+      "loss": 1.1963,
+      "step": 32050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7596234405227965e-05,
+      "loss": 1.208,
+      "step": 32060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.759547275579996e-05,
+      "loss": 1.1889,
+      "step": 32070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.759471110637196e-05,
+      "loss": 1.2029,
+      "step": 32080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.759394945694396e-05,
+      "loss": 1.2693,
+      "step": 32090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.759318780751596e-05,
+      "loss": 1.0761,
+      "step": 32100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.759242615808795e-05,
+      "loss": 1.1896,
+      "step": 32110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7591664508659956e-05,
+      "loss": 1.1595,
+      "step": 32120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7590902859231954e-05,
+      "loss": 1.2456,
+      "step": 32130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.759014120980395e-05,
+      "loss": 1.1452,
+      "step": 32140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758937956037595e-05,
+      "loss": 1.2558,
+      "step": 32150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758861791094795e-05,
+      "loss": 1.296,
+      "step": 32160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758785626151995e-05,
+      "loss": 1.2356,
+      "step": 32170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758709461209195e-05,
+      "loss": 1.0887,
+      "step": 32180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7586332962663945e-05,
+      "loss": 1.3384,
+      "step": 32190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758557131323594e-05,
+      "loss": 1.1962,
+      "step": 32200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758480966380795e-05,
+      "loss": 1.1194,
+      "step": 32210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7584048014379945e-05,
+      "loss": 1.1825,
+      "step": 32220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758328636495194e-05,
+      "loss": 1.1266,
+      "step": 32230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758252471552394e-05,
+      "loss": 1.2856,
+      "step": 32240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7581763066095944e-05,
+      "loss": 1.4114,
+      "step": 32250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7581001416667934e-05,
+      "loss": 1.2799,
+      "step": 32260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.758023976723994e-05,
+      "loss": 1.0213,
+      "step": 32270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7579478117811936e-05,
+      "loss": 1.1963,
+      "step": 32280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7578716468383934e-05,
+      "loss": 1.0686,
+      "step": 32290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.757795481895593e-05,
+      "loss": 1.0671,
+      "step": 32300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7577193169527935e-05,
+      "loss": 1.2896,
+      "step": 32310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7576431520099926e-05,
+      "loss": 1.0697,
+      "step": 32320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.757566987067193e-05,
+      "loss": 1.189,
+      "step": 32330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.757490822124393e-05,
+      "loss": 1.2892,
+      "step": 32340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7574146571815925e-05,
+      "loss": 1.1494,
+      "step": 32350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.757338492238792e-05,
+      "loss": 1.355,
+      "step": 32360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.757262327295993e-05,
+      "loss": 1.2554,
+      "step": 32370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7571861623531924e-05,
+      "loss": 1.1968,
+      "step": 32380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.757109997410392e-05,
+      "loss": 1.1649,
+      "step": 32390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.757033832467592e-05,
+      "loss": 1.2909,
+      "step": 32400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756957667524792e-05,
+      "loss": 1.1755,
+      "step": 32410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756881502581992e-05,
+      "loss": 1.2283,
+      "step": 32420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756805337639192e-05,
+      "loss": 1.2952,
+      "step": 32430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7567291726963916e-05,
+      "loss": 1.2004,
+      "step": 32440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7566530077535913e-05,
+      "loss": 1.154,
+      "step": 32450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756576842810792e-05,
+      "loss": 1.18,
+      "step": 32460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756500677867991e-05,
+      "loss": 1.0639,
+      "step": 32470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756424512925191e-05,
+      "loss": 1.0668,
+      "step": 32480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756348347982391e-05,
+      "loss": 1.2601,
+      "step": 32490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756272183039591e-05,
+      "loss": 1.2438,
+      "step": 32500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7561960180967905e-05,
+      "loss": 1.2868,
+      "step": 32510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.756119853153991e-05,
+      "loss": 1.0897,
+      "step": 32520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75604368821119e-05,
+      "loss": 1.2028,
+      "step": 32530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7559675232683904e-05,
+      "loss": 1.2514,
+      "step": 32540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75589135832559e-05,
+      "loss": 1.3225,
+      "step": 32550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75581519338279e-05,
+      "loss": 1.2561,
+      "step": 32560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7557390284399897e-05,
+      "loss": 1.1143,
+      "step": 32570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7556628634971894e-05,
+      "loss": 0.9515,
+      "step": 32580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75558669855439e-05,
+      "loss": 1.0756,
+      "step": 32590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7555105336115896e-05,
+      "loss": 1.2565,
+      "step": 32600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.755434368668789e-05,
+      "loss": 1.1349,
+      "step": 32610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.755358203725989e-05,
+      "loss": 1.2563,
+      "step": 32620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7552820387831895e-05,
+      "loss": 1.0834,
+      "step": 32630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7552058738403886e-05,
+      "loss": 1.2391,
+      "step": 32640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.755129708897589e-05,
+      "loss": 1.078,
+      "step": 32650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.755053543954789e-05,
+      "loss": 1.2981,
+      "step": 32660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7549773790119885e-05,
+      "loss": 1.1525,
+      "step": 32670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.754901214069188e-05,
+      "loss": 1.184,
+      "step": 32680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7548250491263887e-05,
+      "loss": 1.1183,
+      "step": 32690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.754748884183588e-05,
+      "loss": 1.1868,
+      "step": 32700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.754672719240788e-05,
+      "loss": 1.2233,
+      "step": 32710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.754596554297988e-05,
+      "loss": 1.2463,
+      "step": 32720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7545203893551876e-05,
+      "loss": 1.1219,
+      "step": 32730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7544442244123874e-05,
+      "loss": 1.325,
+      "step": 32740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.754368059469588e-05,
+      "loss": 1.2323,
+      "step": 32750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7542918945267876e-05,
+      "loss": 1.3855,
+      "step": 32760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.754215729583987e-05,
+      "loss": 1.2277,
+      "step": 32770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.754139564641187e-05,
+      "loss": 1.1614,
+      "step": 32780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.754063399698387e-05,
+      "loss": 1.1897,
+      "step": 32790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753987234755587e-05,
+      "loss": 1.1616,
+      "step": 32800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753911069812787e-05,
+      "loss": 1.196,
+      "step": 32810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753834904869987e-05,
+      "loss": 1.2388,
+      "step": 32820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7537587399271865e-05,
+      "loss": 1.1834,
+      "step": 32830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753682574984387e-05,
+      "loss": 1.1587,
+      "step": 32840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753606410041586e-05,
+      "loss": 1.3524,
+      "step": 32850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7535302450987864e-05,
+      "loss": 1.2636,
+      "step": 32860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753454080155986e-05,
+      "loss": 1.1907,
+      "step": 32870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753377915213186e-05,
+      "loss": 1.0947,
+      "step": 32880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7533017502703856e-05,
+      "loss": 1.0717,
+      "step": 32890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753225585327586e-05,
+      "loss": 1.0926,
+      "step": 32900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.753149420384785e-05,
+      "loss": 1.1823,
+      "step": 32910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7530732554419855e-05,
+      "loss": 1.1922,
+      "step": 32920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.752997090499185e-05,
+      "loss": 1.1729,
+      "step": 32930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.752920925556385e-05,
+      "loss": 1.151,
+      "step": 32940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.752844760613585e-05,
+      "loss": 1.2187,
+      "step": 32950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752768595670785e-05,
+      "loss": 1.1405,
+      "step": 32960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752692430727985e-05,
+      "loss": 1.0834,
+      "step": 32970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752616265785185e-05,
+      "loss": 1.0969,
+      "step": 32980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7525401008423844e-05,
+      "loss": 1.1034,
+      "step": 32990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752463935899584e-05,
+      "loss": 1.1276,
+      "step": 33000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7523877709567846e-05,
+      "loss": 1.2579,
+      "step": 33010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752311606013984e-05,
+      "loss": 1.2043,
+      "step": 33020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752235441071184e-05,
+      "loss": 1.1283,
+      "step": 33030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.752159276128384e-05,
+      "loss": 1.1795,
+      "step": 33040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7520831111855836e-05,
+      "loss": 1.0381,
+      "step": 33050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7520069462427833e-05,
+      "loss": 1.2217,
+      "step": 33060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751930781299984e-05,
+      "loss": 1.1314,
+      "step": 33070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751854616357183e-05,
+      "loss": 1.2315,
+      "step": 33080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751778451414383e-05,
+      "loss": 1.2423,
+      "step": 33090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751702286471583e-05,
+      "loss": 1.0333,
+      "step": 33100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751626121528783e-05,
+      "loss": 1.3248,
+      "step": 33110
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7515499565859825e-05,
+      "loss": 1.2045,
+      "step": 33120
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751473791643183e-05,
+      "loss": 1.2317,
+      "step": 33130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751397626700383e-05,
+      "loss": 1.0986,
+      "step": 33140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7513214617575824e-05,
+      "loss": 1.25,
+      "step": 33150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751245296814782e-05,
+      "loss": 1.1812,
+      "step": 33160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.751169131871982e-05,
+      "loss": 1.1247,
+      "step": 33170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.751092966929182e-05,
+      "loss": 1.105,
+      "step": 33180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.751016801986382e-05,
+      "loss": 1.0895,
+      "step": 33190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750940637043582e-05,
+      "loss": 1.1275,
+      "step": 33200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7508644721007816e-05,
+      "loss": 1.1141,
+      "step": 33210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750788307157982e-05,
+      "loss": 1.184,
+      "step": 33220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750712142215181e-05,
+      "loss": 1.2674,
+      "step": 33230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7506359772723815e-05,
+      "loss": 1.1919,
+      "step": 33240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750559812329581e-05,
+      "loss": 1.2431,
+      "step": 33250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750483647386781e-05,
+      "loss": 1.2627,
+      "step": 33260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750407482443981e-05,
+      "loss": 1.1666,
+      "step": 33270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.750331317501181e-05,
+      "loss": 1.0644,
+      "step": 33280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.75025515255838e-05,
+      "loss": 1.115,
+      "step": 33290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7501789876155807e-05,
+      "loss": 1.0781,
+      "step": 33300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7501028226727804e-05,
+      "loss": 1.06,
+      "step": 33310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.75002665772998e-05,
+      "loss": 1.2305,
+      "step": 33320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74995049278718e-05,
+      "loss": 1.2965,
+      "step": 33330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74987432784438e-05,
+      "loss": 1.0925,
+      "step": 33340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74979816290158e-05,
+      "loss": 1.3295,
+      "step": 33350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74972199795878e-05,
+      "loss": 1.1444,
+      "step": 33360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7496458330159796e-05,
+      "loss": 1.18,
+      "step": 33370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.749569668073179e-05,
+      "loss": 1.1992,
+      "step": 33380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74949350313038e-05,
+      "loss": 1.3217,
+      "step": 33390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7494173381875795e-05,
+      "loss": 1.0486,
+      "step": 33400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.749341173244779e-05,
+      "loss": 1.2457,
+      "step": 33410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.749265008301979e-05,
+      "loss": 1.1553,
+      "step": 33420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7491888433591794e-05,
+      "loss": 1.2249,
+      "step": 33430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7491126784163785e-05,
+      "loss": 1.3239,
+      "step": 33440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.749036513473579e-05,
+      "loss": 1.2965,
+      "step": 33450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7489603485307786e-05,
+      "loss": 1.1903,
+      "step": 33460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7488841835879784e-05,
+      "loss": 1.1912,
+      "step": 33470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748808018645178e-05,
+      "loss": 1.1415,
+      "step": 33480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748731853702378e-05,
+      "loss": 1.0928,
+      "step": 33490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7486556887595776e-05,
+      "loss": 1.2203,
+      "step": 33500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748579523816778e-05,
+      "loss": 1.1691,
+      "step": 33510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748503358873978e-05,
+      "loss": 1.227,
+      "step": 33520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7484271939311775e-05,
+      "loss": 1.126,
+      "step": 33530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748351028988377e-05,
+      "loss": 1.1337,
+      "step": 33540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748274864045577e-05,
+      "loss": 1.2202,
+      "step": 33550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7481986991027775e-05,
+      "loss": 1.1453,
+      "step": 33560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748122534159977e-05,
+      "loss": 1.2417,
+      "step": 33570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748046369217177e-05,
+      "loss": 1.1824,
+      "step": 33580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.747970204274377e-05,
+      "loss": 0.9775,
+      "step": 33590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.747894039331577e-05,
+      "loss": 1.2511,
+      "step": 33600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.747817874388776e-05,
+      "loss": 1.1457,
+      "step": 33610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7477417094459766e-05,
+      "loss": 1.2101,
+      "step": 33620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7476655445031764e-05,
+      "loss": 1.2484,
+      "step": 33630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.747589379560376e-05,
+      "loss": 1.1356,
+      "step": 33640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.747513214617576e-05,
+      "loss": 1.1071,
+      "step": 33650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.747437049674776e-05,
+      "loss": 1.1681,
+      "step": 33660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7473608847319753e-05,
+      "loss": 1.1793,
+      "step": 33670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.747284719789176e-05,
+      "loss": 1.157,
+      "step": 33680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7472085548463755e-05,
+      "loss": 1.0583,
+      "step": 33690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.747132389903575e-05,
+      "loss": 1.3381,
+      "step": 33700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.747056224960775e-05,
+      "loss": 1.1891,
+      "step": 33710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7469800600179754e-05,
+      "loss": 1.1772,
+      "step": 33720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.746903895075175e-05,
+      "loss": 1.2132,
+      "step": 33730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.746827730132375e-05,
+      "loss": 1.3305,
+      "step": 33740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.746751565189575e-05,
+      "loss": 1.2697,
+      "step": 33750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7466754002467744e-05,
+      "loss": 1.2098,
+      "step": 33760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.746599235303975e-05,
+      "loss": 1.0728,
+      "step": 33770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7465230703611746e-05,
+      "loss": 1.1236,
+      "step": 33780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7464469054183743e-05,
+      "loss": 1.3067,
+      "step": 33790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.746370740475574e-05,
+      "loss": 1.2175,
+      "step": 33800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7462945755327745e-05,
+      "loss": 1.1536,
+      "step": 33810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7462184105899736e-05,
+      "loss": 1.132,
+      "step": 33820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.746142245647174e-05,
+      "loss": 1.2493,
+      "step": 33830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.746066080704374e-05,
+      "loss": 1.1704,
+      "step": 33840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7459899157615735e-05,
+      "loss": 1.1754,
+      "step": 33850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745913750818773e-05,
+      "loss": 1.1094,
+      "step": 33860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745837585875974e-05,
+      "loss": 1.1484,
+      "step": 33870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745761420933173e-05,
+      "loss": 1.2179,
+      "step": 33880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745685255990373e-05,
+      "loss": 1.0915,
+      "step": 33890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745609091047573e-05,
+      "loss": 1.3777,
+      "step": 33900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7455329261047727e-05,
+      "loss": 1.1434,
+      "step": 33910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7454567611619724e-05,
+      "loss": 1.1441,
+      "step": 33920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745380596219172e-05,
+      "loss": 1.064,
+      "step": 33930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7453044312763726e-05,
+      "loss": 1.0862,
+      "step": 33940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745228266333572e-05,
+      "loss": 1.2215,
+      "step": 33950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745152101390772e-05,
+      "loss": 1.2215,
+      "step": 33960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.745075936447972e-05,
+      "loss": 1.206,
+      "step": 33970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.744999771505172e-05,
+      "loss": 1.2686,
+      "step": 33980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.744923606562371e-05,
+      "loss": 1.2035,
+      "step": 33990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.744847441619572e-05,
+      "loss": 1.1376,
+      "step": 34000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7447712766767715e-05,
+      "loss": 1.199,
+      "step": 34010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.744695111733971e-05,
+      "loss": 1.1153,
+      "step": 34020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.744618946791171e-05,
+      "loss": 1.1323,
+      "step": 34030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7445427818483714e-05,
+      "loss": 1.1458,
+      "step": 34040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7444666169055705e-05,
+      "loss": 1.2376,
+      "step": 34050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.744390451962771e-05,
+      "loss": 1.2842,
+      "step": 34060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7443142870199706e-05,
+      "loss": 1.1047,
+      "step": 34070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7442381220771704e-05,
+      "loss": 1.3179,
+      "step": 34080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74416195713437e-05,
+      "loss": 1.0788,
+      "step": 34090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7440857921915706e-05,
+      "loss": 1.2751,
+      "step": 34100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74400962724877e-05,
+      "loss": 1.2307,
+      "step": 34110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74393346230597e-05,
+      "loss": 1.1677,
+      "step": 34120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74385729736317e-05,
+      "loss": 1.0551,
+      "step": 34130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7437811324203695e-05,
+      "loss": 1.1539,
+      "step": 34140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74370496747757e-05,
+      "loss": 1.1459,
+      "step": 34150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74362880253477e-05,
+      "loss": 1.1516,
+      "step": 34160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7435526375919695e-05,
+      "loss": 1.2353,
+      "step": 34170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.743476472649169e-05,
+      "loss": 1.3092,
+      "step": 34180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7434003077063696e-05,
+      "loss": 1.151,
+      "step": 34190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.743324142763569e-05,
+      "loss": 1.1637,
+      "step": 34200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.743247977820769e-05,
+      "loss": 1.0167,
+      "step": 34210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.743171812877969e-05,
+      "loss": 1.1749,
+      "step": 34220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7430956479351686e-05,
+      "loss": 1.2405,
+      "step": 34230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7430194829923684e-05,
+      "loss": 1.1015,
+      "step": 34240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.742943318049569e-05,
+      "loss": 1.1664,
+      "step": 34250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.742867153106768e-05,
+      "loss": 1.3228,
+      "step": 34260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.742790988163968e-05,
+      "loss": 1.1304,
+      "step": 34270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742714823221168e-05,
+      "loss": 1.2278,
+      "step": 34280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742638658278368e-05,
+      "loss": 1.1638,
+      "step": 34290
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7425624933355675e-05,
+      "loss": 1.351,
+      "step": 34300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742486328392768e-05,
+      "loss": 1.1981,
+      "step": 34310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742410163449968e-05,
+      "loss": 1.1588,
+      "step": 34320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7423339985071674e-05,
+      "loss": 1.1057,
+      "step": 34330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742257833564367e-05,
+      "loss": 1.1507,
+      "step": 34340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742181668621567e-05,
+      "loss": 1.0128,
+      "step": 34350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7421055036787674e-05,
+      "loss": 1.0881,
+      "step": 34360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.742029338735967e-05,
+      "loss": 1.1596,
+      "step": 34370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741953173793167e-05,
+      "loss": 1.0229,
+      "step": 34380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7418770088503666e-05,
+      "loss": 1.1685,
+      "step": 34390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7418008439075663e-05,
+      "loss": 1.1341,
+      "step": 34400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741724678964766e-05,
+      "loss": 1.0274,
+      "step": 34410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7416485140219665e-05,
+      "loss": 1.1388,
+      "step": 34420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7415723490791656e-05,
+      "loss": 1.2147,
+      "step": 34430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741496184136366e-05,
+      "loss": 1.1113,
+      "step": 34440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741420019193566e-05,
+      "loss": 1.2347,
+      "step": 34450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7413438542507655e-05,
+      "loss": 1.1117,
+      "step": 34460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741267689307965e-05,
+      "loss": 1.2386,
+      "step": 34470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741191524365166e-05,
+      "loss": 1.2357,
+      "step": 34480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.741115359422365e-05,
+      "loss": 1.2404,
+      "step": 34490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.741039194479565e-05,
+      "loss": 1.2643,
+      "step": 34500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740963029536765e-05,
+      "loss": 1.0569,
+      "step": 34510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7408868645939647e-05,
+      "loss": 1.2904,
+      "step": 34520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740810699651165e-05,
+      "loss": 1.2174,
+      "step": 34530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740734534708365e-05,
+      "loss": 1.1135,
+      "step": 34540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7406583697655646e-05,
+      "loss": 1.2441,
+      "step": 34550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740582204822764e-05,
+      "loss": 1.2187,
+      "step": 34560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740506039879965e-05,
+      "loss": 1.194,
+      "step": 34570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740429874937164e-05,
+      "loss": 1.1255,
+      "step": 34580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740353709994364e-05,
+      "loss": 1.1232,
+      "step": 34590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740277545051564e-05,
+      "loss": 1.1251,
+      "step": 34600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740201380108764e-05,
+      "loss": 1.3046,
+      "step": 34610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7401252151659635e-05,
+      "loss": 1.052,
+      "step": 34620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.740049050223164e-05,
+      "loss": 1.2359,
+      "step": 34630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.739972885280363e-05,
+      "loss": 1.226,
+      "step": 34640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7398967203375634e-05,
+      "loss": 1.138,
+      "step": 34650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.739820555394763e-05,
+      "loss": 1.203,
+      "step": 34660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.739744390451963e-05,
+      "loss": 1.2686,
+      "step": 34670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7396682255091626e-05,
+      "loss": 1.1672,
+      "step": 34680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.739592060566363e-05,
+      "loss": 1.2457,
+      "step": 34690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.739515895623562e-05,
+      "loss": 1.2471,
+      "step": 34700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7394397306807626e-05,
+      "loss": 1.2718,
+      "step": 34710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.739363565737962e-05,
+      "loss": 1.143,
+      "step": 34720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.739287400795162e-05,
+      "loss": 1.123,
+      "step": 34730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7392112358523625e-05,
+      "loss": 1.0053,
+      "step": 34740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.739135070909562e-05,
+      "loss": 1.1769,
+      "step": 34750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.739058905966762e-05,
+      "loss": 1.2503,
+      "step": 34760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738982741023962e-05,
+      "loss": 1.0207,
+      "step": 34770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738906576081162e-05,
+      "loss": 1.2039,
+      "step": 34780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738830411138361e-05,
+      "loss": 1.2071,
+      "step": 34790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7387542461955616e-05,
+      "loss": 1.2963,
+      "step": 34800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7386780812527614e-05,
+      "loss": 1.1788,
+      "step": 34810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738601916309961e-05,
+      "loss": 1.1589,
+      "step": 34820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738525751367161e-05,
+      "loss": 1.2658,
+      "step": 34830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738449586424361e-05,
+      "loss": 1.0678,
+      "step": 34840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7383734214815604e-05,
+      "loss": 1.1374,
+      "step": 34850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.738297256538761e-05,
+      "loss": 1.2535,
+      "step": 34860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73822109159596e-05,
+      "loss": 1.1879,
+      "step": 34870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73814492665316e-05,
+      "loss": 1.2196,
+      "step": 34880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73806876171036e-05,
+      "loss": 1.2217,
+      "step": 34890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73799259676756e-05,
+      "loss": 1.22,
+      "step": 34900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73791643182476e-05,
+      "loss": 1.192,
+      "step": 34910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73784026688196e-05,
+      "loss": 1.1269,
+      "step": 34920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73776410193916e-05,
+      "loss": 1.155,
+      "step": 34930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7376879369963594e-05,
+      "loss": 1.1453,
+      "step": 34940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.73761177205356e-05,
+      "loss": 1.2234,
+      "step": 34950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.737535607110759e-05,
+      "loss": 1.3529,
+      "step": 34960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7374594421679594e-05,
+      "loss": 1.2162,
+      "step": 34970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.737383277225159e-05,
+      "loss": 1.1344,
+      "step": 34980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.737307112282359e-05,
+      "loss": 1.266,
+      "step": 34990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7372309473395586e-05,
+      "loss": 1.1097,
+      "step": 35000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.737154782396759e-05,
+      "loss": 1.1934,
+      "step": 35010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.737078617453958e-05,
+      "loss": 1.2184,
+      "step": 35020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7370024525111585e-05,
+      "loss": 1.1424,
+      "step": 35030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736926287568358e-05,
+      "loss": 1.1326,
+      "step": 35040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736850122625558e-05,
+      "loss": 1.1821,
+      "step": 35050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736773957682758e-05,
+      "loss": 1.1228,
+      "step": 35060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736697792739958e-05,
+      "loss": 1.1651,
+      "step": 35070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736621627797157e-05,
+      "loss": 1.1536,
+      "step": 35080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736545462854358e-05,
+      "loss": 1.1886,
+      "step": 35090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7364692979115574e-05,
+      "loss": 1.1076,
+      "step": 35100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736393132968757e-05,
+      "loss": 1.1829,
+      "step": 35110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7363169680259576e-05,
+      "loss": 1.0728,
+      "step": 35120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736240803083157e-05,
+      "loss": 1.2863,
+      "step": 35130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.736164638140357e-05,
+      "loss": 1.1869,
+      "step": 35140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.736088473197557e-05,
+      "loss": 1.1725,
+      "step": 35150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.736012308254757e-05,
+      "loss": 1.1048,
+      "step": 35160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.735936143311956e-05,
+      "loss": 1.1374,
+      "step": 35170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.735859978369157e-05,
+      "loss": 1.1588,
+      "step": 35180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7357838134263565e-05,
+      "loss": 1.3128,
+      "step": 35190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.735707648483556e-05,
+      "loss": 1.1474,
+      "step": 35200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.735631483540756e-05,
+      "loss": 1.2077,
+      "step": 35210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7355553185979564e-05,
+      "loss": 1.1394,
+      "step": 35220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7354791536551555e-05,
+      "loss": 1.0278,
+      "step": 35230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.735402988712356e-05,
+      "loss": 1.0988,
+      "step": 35240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7353268237695556e-05,
+      "loss": 1.0682,
+      "step": 35250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7352506588267554e-05,
+      "loss": 1.099,
+      "step": 35260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.735174493883955e-05,
+      "loss": 1.2213,
+      "step": 35270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7350983289411556e-05,
+      "loss": 1.1745,
+      "step": 35280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7350221639983546e-05,
+      "loss": 1.0298,
+      "step": 35290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.734945999055555e-05,
+      "loss": 1.0471,
+      "step": 35300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.734869834112755e-05,
+      "loss": 1.1571,
+      "step": 35310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7347936691699546e-05,
+      "loss": 1.0954,
+      "step": 35320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.734717504227155e-05,
+      "loss": 1.2435,
+      "step": 35330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.734641339284354e-05,
+      "loss": 1.0657,
+      "step": 35340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7345651743415545e-05,
+      "loss": 1.1303,
+      "step": 35350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.734489009398754e-05,
+      "loss": 0.9784,
+      "step": 35360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.734412844455954e-05,
+      "loss": 0.9903,
+      "step": 35370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.734336679513154e-05,
+      "loss": 1.2182,
+      "step": 35380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.734260514570354e-05,
+      "loss": 1.1138,
+      "step": 35390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.734184349627553e-05,
+      "loss": 1.0311,
+      "step": 35400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7341081846847536e-05,
+      "loss": 1.0199,
+      "step": 35410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7340320197419534e-05,
+      "loss": 1.2857,
+      "step": 35420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733955854799153e-05,
+      "loss": 1.0777,
+      "step": 35430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733879689856353e-05,
+      "loss": 1.0834,
+      "step": 35440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733803524913553e-05,
+      "loss": 1.1155,
+      "step": 35450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7337273599707524e-05,
+      "loss": 1.1016,
+      "step": 35460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733651195027953e-05,
+      "loss": 1.3351,
+      "step": 35470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7335750300851525e-05,
+      "loss": 1.2441,
+      "step": 35480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733498865142352e-05,
+      "loss": 1.1316,
+      "step": 35490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733422700199552e-05,
+      "loss": 1.1574,
+      "step": 35500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7333465352567525e-05,
+      "loss": 1.1625,
+      "step": 35510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733270370313952e-05,
+      "loss": 1.2165,
+      "step": 35520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.733194205371152e-05,
+      "loss": 1.1387,
+      "step": 35530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7331180404283524e-05,
+      "loss": 1.1616,
+      "step": 35540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7330418754855514e-05,
+      "loss": 1.1878,
+      "step": 35550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.732965710542752e-05,
+      "loss": 1.1038,
+      "step": 35560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7328895455999516e-05,
+      "loss": 1.1517,
+      "step": 35570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7328133806571514e-05,
+      "loss": 1.0822,
+      "step": 35580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.732737215714351e-05,
+      "loss": 1.2437,
+      "step": 35590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7326610507715515e-05,
+      "loss": 1.0899,
+      "step": 35600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7325848858287506e-05,
+      "loss": 1.2006,
+      "step": 35610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.732508720885951e-05,
+      "loss": 1.0941,
+      "step": 35620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.732432555943151e-05,
+      "loss": 1.057,
+      "step": 35630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7323563910003505e-05,
+      "loss": 1.1536,
+      "step": 35640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73228022605755e-05,
+      "loss": 1.3341,
+      "step": 35650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.732204061114751e-05,
+      "loss": 1.3301,
+      "step": 35660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73212789617195e-05,
+      "loss": 1.2434,
+      "step": 35670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73205173122915e-05,
+      "loss": 1.1467,
+      "step": 35680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73197556628635e-05,
+      "loss": 1.1344,
+      "step": 35690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73189940134355e-05,
+      "loss": 1.0021,
+      "step": 35700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73182323640075e-05,
+      "loss": 1.074,
+      "step": 35710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73174707145795e-05,
+      "loss": 1.1477,
+      "step": 35720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7316709065151496e-05,
+      "loss": 0.9624,
+      "step": 35730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.731594741572349e-05,
+      "loss": 1.1444,
+      "step": 35740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.73151857662955e-05,
+      "loss": 1.219,
+      "step": 35750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.731442411686749e-05,
+      "loss": 1.0799,
+      "step": 35760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.731366246743949e-05,
+      "loss": 1.1945,
+      "step": 35770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.731290081801148e-05,
+      "loss": 1.1531,
+      "step": 35780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.731213916858349e-05,
+      "loss": 1.1395,
+      "step": 35790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7311377519155485e-05,
+      "loss": 1.2338,
+      "step": 35800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.731061586972748e-05,
+      "loss": 1.191,
+      "step": 35810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.730985422029948e-05,
+      "loss": 1.1575,
+      "step": 35820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7309092570871484e-05,
+      "loss": 1.2275,
+      "step": 35830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7308330921443475e-05,
+      "loss": 1.2614,
+      "step": 35840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.730756927201548e-05,
+      "loss": 1.1963,
+      "step": 35850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7306807622587477e-05,
+      "loss": 1.1298,
+      "step": 35860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7306045973159474e-05,
+      "loss": 1.1342,
+      "step": 35870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.730528432373147e-05,
+      "loss": 0.9548,
+      "step": 35880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7304522674303476e-05,
+      "loss": 1.1643,
+      "step": 35890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.730376102487547e-05,
+      "loss": 1.2084,
+      "step": 35900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.730299937544747e-05,
+      "loss": 1.1802,
+      "step": 35910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7302237726019475e-05,
+      "loss": 1.18,
+      "step": 35920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7301476076591466e-05,
+      "loss": 1.1982,
+      "step": 35930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.730071442716347e-05,
+      "loss": 1.2347,
+      "step": 35940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.729995277773547e-05,
+      "loss": 1.1243,
+      "step": 35950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7299191128307465e-05,
+      "loss": 1.303,
+      "step": 35960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.729842947887946e-05,
+      "loss": 1.2121,
+      "step": 35970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7297667829451466e-05,
+      "loss": 1.2737,
+      "step": 35980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.729690618002346e-05,
+      "loss": 1.2268,
+      "step": 35990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.729614453059546e-05,
+      "loss": 1.3532,
+      "step": 36000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.729538288116746e-05,
+      "loss": 1.3012,
+      "step": 36010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7294621231739456e-05,
+      "loss": 1.3511,
+      "step": 36020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7293859582311454e-05,
+      "loss": 1.1769,
+      "step": 36030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.729309793288346e-05,
+      "loss": 1.2012,
+      "step": 36040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.729233628345545e-05,
+      "loss": 1.0515,
+      "step": 36050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.729157463402745e-05,
+      "loss": 1.1662,
+      "step": 36060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.729081298459945e-05,
+      "loss": 1.2,
+      "step": 36070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.729005133517145e-05,
+      "loss": 1.1139,
+      "step": 36080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7289289685743445e-05,
+      "loss": 1.134,
+      "step": 36090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728852803631545e-05,
+      "loss": 1.0821,
+      "step": 36100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728776638688745e-05,
+      "loss": 1.1025,
+      "step": 36110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7287004737459445e-05,
+      "loss": 1.0847,
+      "step": 36120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728624308803145e-05,
+      "loss": 1.2654,
+      "step": 36130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728548143860344e-05,
+      "loss": 1.1584,
+      "step": 36140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7284719789175444e-05,
+      "loss": 1.0561,
+      "step": 36150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728395813974744e-05,
+      "loss": 1.0982,
+      "step": 36160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728319649031944e-05,
+      "loss": 1.0464,
+      "step": 36170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7282434840891436e-05,
+      "loss": 1.2221,
+      "step": 36180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728167319146344e-05,
+      "loss": 1.1138,
+      "step": 36190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.728091154203543e-05,
+      "loss": 1.0956,
+      "step": 36200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7280149892607435e-05,
+      "loss": 1.1566,
+      "step": 36210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.727938824317943e-05,
+      "loss": 1.1652,
+      "step": 36220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.727862659375143e-05,
+      "loss": 1.2867,
+      "step": 36230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.727786494432343e-05,
+      "loss": 1.1401,
+      "step": 36240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7277103294895425e-05,
+      "loss": 1.1544,
+      "step": 36250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727634164546742e-05,
+      "loss": 1.1776,
+      "step": 36260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727557999603943e-05,
+      "loss": 1.2891,
+      "step": 36270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7274818346611424e-05,
+      "loss": 1.0189,
+      "step": 36280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727405669718342e-05,
+      "loss": 1.2551,
+      "step": 36290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727329504775542e-05,
+      "loss": 1.0859,
+      "step": 36300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727253339832742e-05,
+      "loss": 1.0938,
+      "step": 36310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727177174889942e-05,
+      "loss": 1.1616,
+      "step": 36320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.727101009947142e-05,
+      "loss": 1.1536,
+      "step": 36330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7270248450043416e-05,
+      "loss": 1.2209,
+      "step": 36340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726948680061541e-05,
+      "loss": 1.1691,
+      "step": 36350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726872515118742e-05,
+      "loss": 1.1108,
+      "step": 36360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726796350175941e-05,
+      "loss": 1.2335,
+      "step": 36370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726720185233141e-05,
+      "loss": 1.1459,
+      "step": 36380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726644020290341e-05,
+      "loss": 1.2391,
+      "step": 36390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726567855347541e-05,
+      "loss": 1.2256,
+      "step": 36400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7264916904047405e-05,
+      "loss": 1.0692,
+      "step": 36410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.726415525461941e-05,
+      "loss": 1.0307,
+      "step": 36420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.72633936051914e-05,
+      "loss": 1.1629,
+      "step": 36430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7262631955763404e-05,
+      "loss": 1.151,
+      "step": 36440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.72618703063354e-05,
+      "loss": 1.1355,
+      "step": 36450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.72611086569074e-05,
+      "loss": 1.0513,
+      "step": 36460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7260347007479397e-05,
+      "loss": 1.0162,
+      "step": 36470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.72595853580514e-05,
+      "loss": 1.2383,
+      "step": 36480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.72588237086234e-05,
+      "loss": 1.0632,
+      "step": 36490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7258062059195396e-05,
+      "loss": 1.1571,
+      "step": 36500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.72573004097674e-05,
+      "loss": 1.1409,
+      "step": 36510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.725653876033939e-05,
+      "loss": 1.2554,
+      "step": 36520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7255777110911395e-05,
+      "loss": 0.9508,
+      "step": 36530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.725501546148339e-05,
+      "loss": 1.2951,
+      "step": 36540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.725425381205539e-05,
+      "loss": 1.1654,
+      "step": 36550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.725349216262739e-05,
+      "loss": 1.2309,
+      "step": 36560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.725273051319939e-05,
+      "loss": 1.16,
+      "step": 36570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.725196886377138e-05,
+      "loss": 1.0789,
+      "step": 36580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7251207214343386e-05,
+      "loss": 1.2275,
+      "step": 36590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7250445564915384e-05,
+      "loss": 1.1138,
+      "step": 36600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.724968391548738e-05,
+      "loss": 1.1847,
+      "step": 36610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.724892226605938e-05,
+      "loss": 1.1492,
+      "step": 36620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.724816061663138e-05,
+      "loss": 1.1946,
+      "step": 36630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7247398967203374e-05,
+      "loss": 1.0755,
+      "step": 36640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.724663731777538e-05,
+      "loss": 1.1115,
+      "step": 36650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7245875668347375e-05,
+      "loss": 1.0872,
+      "step": 36660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.724511401891937e-05,
+      "loss": 1.1207,
+      "step": 36670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.724435236949137e-05,
+      "loss": 1.1229,
+      "step": 36680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.724359072006337e-05,
+      "loss": 1.0678,
+      "step": 36690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.724282907063537e-05,
+      "loss": 1.1056,
+      "step": 36700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.724206742120737e-05,
+      "loss": 1.194,
+      "step": 36710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.724130577177937e-05,
+      "loss": 1.1936,
+      "step": 36720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7240544122351365e-05,
+      "loss": 1.1546,
+      "step": 36730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723978247292337e-05,
+      "loss": 1.2247,
+      "step": 36740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723902082349536e-05,
+      "loss": 1.1022,
+      "step": 36750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7238259174067364e-05,
+      "loss": 1.1841,
+      "step": 36760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723749752463936e-05,
+      "loss": 1.2327,
+      "step": 36770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723673587521136e-05,
+      "loss": 1.1094,
+      "step": 36780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7235974225783356e-05,
+      "loss": 1.1389,
+      "step": 36790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723521257635536e-05,
+      "loss": 1.0504,
+      "step": 36800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723445092692735e-05,
+      "loss": 1.0369,
+      "step": 36810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7233689277499355e-05,
+      "loss": 1.111,
+      "step": 36820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723292762807135e-05,
+      "loss": 1.0453,
+      "step": 36830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723216597864335e-05,
+      "loss": 1.1408,
+      "step": 36840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723140432921535e-05,
+      "loss": 1.3026,
+      "step": 36850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.723064267978735e-05,
+      "loss": 1.0856,
+      "step": 36860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.722988103035935e-05,
+      "loss": 1.0908,
+      "step": 36870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.722911938093135e-05,
+      "loss": 1.1789,
+      "step": 36880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7228357731503344e-05,
+      "loss": 1.1513,
+      "step": 36890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.722759608207534e-05,
+      "loss": 1.131,
+      "step": 36900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7226834432647346e-05,
+      "loss": 1.1233,
+      "step": 36910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7226072783219344e-05,
+      "loss": 1.3406,
+      "step": 36920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722531113379134e-05,
+      "loss": 1.0975,
+      "step": 36930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722454948436334e-05,
+      "loss": 1.2286,
+      "step": 36940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722378783493534e-05,
+      "loss": 1.0977,
+      "step": 36950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722302618550733e-05,
+      "loss": 1.1401,
+      "step": 36960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722226453607934e-05,
+      "loss": 1.0859,
+      "step": 36970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7221502886651335e-05,
+      "loss": 1.2538,
+      "step": 36980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722074123722333e-05,
+      "loss": 1.1302,
+      "step": 36990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721997958779533e-05,
+      "loss": 1.0842,
+      "step": 37000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7219217938367334e-05,
+      "loss": 1.0082,
+      "step": 37010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7218456288939325e-05,
+      "loss": 1.1638,
+      "step": 37020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721769463951133e-05,
+      "loss": 1.2538,
+      "step": 37030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721693299008333e-05,
+      "loss": 1.3573,
+      "step": 37040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7216171340655324e-05,
+      "loss": 1.1323,
+      "step": 37050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721540969122732e-05,
+      "loss": 1.1868,
+      "step": 37060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7214648041799326e-05,
+      "loss": 1.1035,
+      "step": 37070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721388639237132e-05,
+      "loss": 1.2239,
+      "step": 37080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721312474294332e-05,
+      "loss": 1.1689,
+      "step": 37090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721236309351532e-05,
+      "loss": 1.1653,
+      "step": 37100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7211601444087316e-05,
+      "loss": 1.1416,
+      "step": 37110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721083979465932e-05,
+      "loss": 1.1239,
+      "step": 37120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.721007814523132e-05,
+      "loss": 1.2078,
+      "step": 37130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7209316495803315e-05,
+      "loss": 1.1802,
+      "step": 37140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.720855484637531e-05,
+      "loss": 1.1882,
+      "step": 37150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.720779319694731e-05,
+      "loss": 1.1271,
+      "step": 37160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.720703154751931e-05,
+      "loss": 1.1899,
+      "step": 37170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.720626989809131e-05,
+      "loss": 1.1236,
+      "step": 37180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.72055082486633e-05,
+      "loss": 1.0945,
+      "step": 37190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7204746599235306e-05,
+      "loss": 1.1023,
+      "step": 37200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7203984949807304e-05,
+      "loss": 1.0701,
+      "step": 37210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.72032233003793e-05,
+      "loss": 1.1411,
+      "step": 37220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.72024616509513e-05,
+      "loss": 1.3111,
+      "step": 37230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.72017000015233e-05,
+      "loss": 1.2272,
+      "step": 37240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.72009383520953e-05,
+      "loss": 1.2205,
+      "step": 37250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.72001767026673e-05,
+      "loss": 1.1312,
+      "step": 37260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7199415053239296e-05,
+      "loss": 1.0974,
+      "step": 37270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.719865340381129e-05,
+      "loss": 1.1408,
+      "step": 37280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.71978917543833e-05,
+      "loss": 1.2198,
+      "step": 37290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7197130104955295e-05,
+      "loss": 1.1775,
+      "step": 37300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.719636845552729e-05,
+      "loss": 1.087,
+      "step": 37310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.719560680609929e-05,
+      "loss": 1.2222,
+      "step": 37320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7194845156671294e-05,
+      "loss": 1.1028,
+      "step": 37330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7194083507243285e-05,
+      "loss": 1.132,
+      "step": 37340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.719332185781529e-05,
+      "loss": 1.1349,
+      "step": 37350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7192560208387286e-05,
+      "loss": 1.1694,
+      "step": 37360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7191798558959284e-05,
+      "loss": 1.1963,
+      "step": 37370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.719103690953128e-05,
+      "loss": 1.1034,
+      "step": 37380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7190275260103285e-05,
+      "loss": 1.1309,
+      "step": 37390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7189513610675276e-05,
+      "loss": 1.2188,
+      "step": 37400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718875196124728e-05,
+      "loss": 1.1483,
+      "step": 37410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718799031181928e-05,
+      "loss": 1.1746,
+      "step": 37420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7187228662391275e-05,
+      "loss": 1.2303,
+      "step": 37430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718646701296327e-05,
+      "loss": 1.2971,
+      "step": 37440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718570536353528e-05,
+      "loss": 1.0934,
+      "step": 37450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7184943714107274e-05,
+      "loss": 1.105,
+      "step": 37460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718418206467927e-05,
+      "loss": 1.2197,
+      "step": 37470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718342041525127e-05,
+      "loss": 1.243,
+      "step": 37480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718265876582327e-05,
+      "loss": 1.0762,
+      "step": 37490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718189711639527e-05,
+      "loss": 1.1113,
+      "step": 37500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.718113546696727e-05,
+      "loss": 0.9767,
+      "step": 37510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7180373817539266e-05,
+      "loss": 1.1991,
+      "step": 37520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7179612168111264e-05,
+      "loss": 1.2268,
+      "step": 37530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.717885051868327e-05,
+      "loss": 1.2455,
+      "step": 37540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.717808886925526e-05,
+      "loss": 1.2276,
+      "step": 37550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717732721982726e-05,
+      "loss": 1.2878,
+      "step": 37560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717656557039926e-05,
+      "loss": 1.1518,
+      "step": 37570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717580392097126e-05,
+      "loss": 1.2156,
+      "step": 37580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7175042271543255e-05,
+      "loss": 1.1456,
+      "step": 37590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717428062211526e-05,
+      "loss": 1.1421,
+      "step": 37600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717351897268725e-05,
+      "loss": 1.1359,
+      "step": 37610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7172757323259254e-05,
+      "loss": 1.2253,
+      "step": 37620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717199567383125e-05,
+      "loss": 1.2887,
+      "step": 37630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717123402440325e-05,
+      "loss": 1.128,
+      "step": 37640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717047237497525e-05,
+      "loss": 1.1511,
+      "step": 37650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7169710725547244e-05,
+      "loss": 1.2705,
+      "step": 37660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.716894907611925e-05,
+      "loss": 1.2132,
+      "step": 37670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7168187426691246e-05,
+      "loss": 1.0374,
+      "step": 37680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.716742577726324e-05,
+      "loss": 1.1153,
+      "step": 37690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.716666412783524e-05,
+      "loss": 1.1278,
+      "step": 37700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7165902478407245e-05,
+      "loss": 1.2323,
+      "step": 37710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7165140828979236e-05,
+      "loss": 1.3125,
+      "step": 37720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.716437917955124e-05,
+      "loss": 1.1728,
+      "step": 37730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.716361753012324e-05,
+      "loss": 1.1913,
+      "step": 37740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7162855880695235e-05,
+      "loss": 1.1608,
+      "step": 37750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.716209423126723e-05,
+      "loss": 1.114,
+      "step": 37760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7161332581839237e-05,
+      "loss": 1.1998,
+      "step": 37770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.716057093241123e-05,
+      "loss": 1.2241,
+      "step": 37780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715980928298323e-05,
+      "loss": 1.1284,
+      "step": 37790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715904763355523e-05,
+      "loss": 1.1232,
+      "step": 37800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7158285984127226e-05,
+      "loss": 1.1888,
+      "step": 37810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7157524334699224e-05,
+      "loss": 1.1936,
+      "step": 37820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715676268527123e-05,
+      "loss": 1.131,
+      "step": 37830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7156001035843226e-05,
+      "loss": 1.1182,
+      "step": 37840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715523938641522e-05,
+      "loss": 1.0133,
+      "step": 37850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715447773698722e-05,
+      "loss": 1.2148,
+      "step": 37860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715371608755922e-05,
+      "loss": 1.2768,
+      "step": 37870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715295443813122e-05,
+      "loss": 1.192,
+      "step": 37880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715219278870322e-05,
+      "loss": 1.1016,
+      "step": 37890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.715143113927522e-05,
+      "loss": 1.0841,
+      "step": 37900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7150669489847215e-05,
+      "loss": 1.1487,
+      "step": 37910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.714990784041922e-05,
+      "loss": 1.0795,
+      "step": 37920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.714914619099121e-05,
+      "loss": 1.1437,
+      "step": 37930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7148384541563214e-05,
+      "loss": 1.1124,
+      "step": 37940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.714762289213521e-05,
+      "loss": 1.1479,
+      "step": 37950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.714686124270721e-05,
+      "loss": 1.1189,
+      "step": 37960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7146099593279206e-05,
+      "loss": 1.1029,
+      "step": 37970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.714533794385121e-05,
+      "loss": 1.2708,
+      "step": 37980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.71445762944232e-05,
+      "loss": 1.2445,
+      "step": 37990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7143814644995205e-05,
+      "loss": 1.1294,
+      "step": 38000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71430529955672e-05,
+      "loss": 1.3055,
+      "step": 38010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71422913461392e-05,
+      "loss": 0.9989,
+      "step": 38020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71415296967112e-05,
+      "loss": 1.1372,
+      "step": 38030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71407680472832e-05,
+      "loss": 1.2659,
+      "step": 38040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71400063978552e-05,
+      "loss": 1.069,
+      "step": 38050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71392447484272e-05,
+      "loss": 1.1199,
+      "step": 38060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7138483098999194e-05,
+      "loss": 1.0738,
+      "step": 38070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713772144957119e-05,
+      "loss": 1.1539,
+      "step": 38080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7136959800143196e-05,
+      "loss": 1.1632,
+      "step": 38090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713619815071519e-05,
+      "loss": 1.1852,
+      "step": 38100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713543650128719e-05,
+      "loss": 1.1288,
+      "step": 38110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713467485185919e-05,
+      "loss": 1.3368,
+      "step": 38120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7133913202431186e-05,
+      "loss": 1.1759,
+      "step": 38130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7133151553003184e-05,
+      "loss": 1.1084,
+      "step": 38140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713238990357519e-05,
+      "loss": 1.0767,
+      "step": 38150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713162825414718e-05,
+      "loss": 1.1217,
+      "step": 38160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713086660471918e-05,
+      "loss": 1.1334,
+      "step": 38170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.713010495529118e-05,
+      "loss": 1.0897,
+      "step": 38180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.712934330586318e-05,
+      "loss": 1.1174,
+      "step": 38190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7128581656435175e-05,
+      "loss": 1.1741,
+      "step": 38200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.712782000700718e-05,
+      "loss": 1.087,
+      "step": 38210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712705835757918e-05,
+      "loss": 1.0205,
+      "step": 38220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7126296708151174e-05,
+      "loss": 1.1188,
+      "step": 38230
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712553505872317e-05,
+      "loss": 1.2671,
+      "step": 38240
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712477340929517e-05,
+      "loss": 1.1042,
+      "step": 38250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7124011759867173e-05,
+      "loss": 1.2004,
+      "step": 38260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712325011043917e-05,
+      "loss": 1.2077,
+      "step": 38270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712248846101117e-05,
+      "loss": 1.1408,
+      "step": 38280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7121726811583166e-05,
+      "loss": 1.1237,
+      "step": 38290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712096516215517e-05,
+      "loss": 1.0108,
+      "step": 38300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.712020351272716e-05,
+      "loss": 1.0229,
+      "step": 38310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7119441863299165e-05,
+      "loss": 1.1368,
+      "step": 38320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711868021387116e-05,
+      "loss": 1.1623,
+      "step": 38330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711791856444316e-05,
+      "loss": 1.2575,
+      "step": 38340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711715691501516e-05,
+      "loss": 1.244,
+      "step": 38350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711639526558716e-05,
+      "loss": 1.0477,
+      "step": 38360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711563361615915e-05,
+      "loss": 1.0622,
+      "step": 38370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711487196673116e-05,
+      "loss": 1.1863,
+      "step": 38380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7114110317303154e-05,
+      "loss": 1.1692,
+      "step": 38390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711334866787515e-05,
+      "loss": 1.2376,
+      "step": 38400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711258701844715e-05,
+      "loss": 1.1662,
+      "step": 38410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711182536901915e-05,
+      "loss": 1.1332,
+      "step": 38420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.711106371959115e-05,
+      "loss": 1.1511,
+      "step": 38430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.711030207016315e-05,
+      "loss": 1.079,
+      "step": 38440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7109540420735146e-05,
+      "loss": 1.2884,
+      "step": 38450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710877877130714e-05,
+      "loss": 1.0948,
+      "step": 38460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710801712187915e-05,
+      "loss": 1.1479,
+      "step": 38470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7107255472451145e-05,
+      "loss": 1.1794,
+      "step": 38480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710649382302314e-05,
+      "loss": 1.1876,
+      "step": 38490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710573217359514e-05,
+      "loss": 1.326,
+      "step": 38500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7104970524167144e-05,
+      "loss": 1.1991,
+      "step": 38510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7104208874739135e-05,
+      "loss": 1.153,
+      "step": 38520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710344722531114e-05,
+      "loss": 1.1169,
+      "step": 38530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710268557588313e-05,
+      "loss": 1.2463,
+      "step": 38540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7101923926455134e-05,
+      "loss": 1.0757,
+      "step": 38550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710116227702713e-05,
+      "loss": 1.0581,
+      "step": 38560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.710040062759913e-05,
+      "loss": 1.0007,
+      "step": 38570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7099638978171126e-05,
+      "loss": 1.039,
+      "step": 38580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.709887732874313e-05,
+      "loss": 1.1078,
+      "step": 38590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.709811567931512e-05,
+      "loss": 1.1689,
+      "step": 38600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7097354029887125e-05,
+      "loss": 1.1651,
+      "step": 38610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.709659238045912e-05,
+      "loss": 1.0866,
+      "step": 38620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.709583073103112e-05,
+      "loss": 1.0837,
+      "step": 38630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7095069081603125e-05,
+      "loss": 1.2186,
+      "step": 38640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.709430743217512e-05,
+      "loss": 1.2137,
+      "step": 38650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.709354578274712e-05,
+      "loss": 1.0412,
+      "step": 38660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.709278413331912e-05,
+      "loss": 1.1294,
+      "step": 38670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.709202248389112e-05,
+      "loss": 1.1169,
+      "step": 38680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.709126083446311e-05,
+      "loss": 1.2643,
+      "step": 38690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7090499185035116e-05,
+      "loss": 1.3671,
+      "step": 38700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7089737535607114e-05,
+      "loss": 1.1035,
+      "step": 38710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.708897588617911e-05,
+      "loss": 1.1137,
+      "step": 38720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.708821423675111e-05,
+      "loss": 1.1617,
+      "step": 38730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.708745258732311e-05,
+      "loss": 1.0669,
+      "step": 38740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7086690937895104e-05,
+      "loss": 1.1425,
+      "step": 38750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.708592928846711e-05,
+      "loss": 1.0864,
+      "step": 38760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7085167639039105e-05,
+      "loss": 1.1695,
+      "step": 38770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70844059896111e-05,
+      "loss": 1.2224,
+      "step": 38780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70836443401831e-05,
+      "loss": 1.0774,
+      "step": 38790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7082882690755104e-05,
+      "loss": 1.1791,
+      "step": 38800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70821210413271e-05,
+      "loss": 1.1576,
+      "step": 38810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70813593918991e-05,
+      "loss": 1.1795,
+      "step": 38820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70805977424711e-05,
+      "loss": 1.1026,
+      "step": 38830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7079836093043094e-05,
+      "loss": 1.1092,
+      "step": 38840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70790744436151e-05,
+      "loss": 1.1113,
+      "step": 38850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7078312794187096e-05,
+      "loss": 1.1662,
+      "step": 38860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7077551144759093e-05,
+      "loss": 1.0589,
+      "step": 38870
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.707678949533109e-05,
+      "loss": 1.0255,
+      "step": 38880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7076027845903095e-05,
+      "loss": 1.105,
+      "step": 38890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7075266196475086e-05,
+      "loss": 1.029,
+      "step": 38900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.707450454704709e-05,
+      "loss": 1.1022,
+      "step": 38910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.707374289761909e-05,
+      "loss": 1.0987,
+      "step": 38920
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7072981248191085e-05,
+      "loss": 1.3199,
+      "step": 38930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.707221959876308e-05,
+      "loss": 1.1216,
+      "step": 38940
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.707145794933509e-05,
+      "loss": 1.1994,
+      "step": 38950
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.707069629990708e-05,
+      "loss": 1.1357,
+      "step": 38960
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706993465047908e-05,
+      "loss": 1.1495,
+      "step": 38970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706917300105108e-05,
+      "loss": 1.1145,
+      "step": 38980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706841135162308e-05,
+      "loss": 1.2672,
+      "step": 38990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7067649702195074e-05,
+      "loss": 1.1278,
+      "step": 39000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706688805276707e-05,
+      "loss": 1.2244,
+      "step": 39010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7066126403339076e-05,
+      "loss": 1.1201,
+      "step": 39020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706536475391107e-05,
+      "loss": 1.1107,
+      "step": 39030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706460310448307e-05,
+      "loss": 1.2221,
+      "step": 39040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706384145505507e-05,
+      "loss": 1.1778,
+      "step": 39050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706307980562707e-05,
+      "loss": 1.2135,
+      "step": 39060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706231815619906e-05,
+      "loss": 1.2152,
+      "step": 39070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.706155650677107e-05,
+      "loss": 0.9721,
+      "step": 39080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7060794857343065e-05,
+      "loss": 1.1544,
+      "step": 39090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.706003320791506e-05,
+      "loss": 1.1458,
+      "step": 39100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.705927155848706e-05,
+      "loss": 1.1541,
+      "step": 39110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7058509909059064e-05,
+      "loss": 1.1074,
+      "step": 39120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7057748259631055e-05,
+      "loss": 1.1938,
+      "step": 39130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.705698661020306e-05,
+      "loss": 1.2906,
+      "step": 39140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7056224960775056e-05,
+      "loss": 1.316,
+      "step": 39150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7055463311347054e-05,
+      "loss": 1.1853,
+      "step": 39160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.705470166191905e-05,
+      "loss": 1.1945,
+      "step": 39170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7053940012491056e-05,
+      "loss": 1.159,
+      "step": 39180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7053178363063046e-05,
+      "loss": 1.2257,
+      "step": 39190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.705241671363505e-05,
+      "loss": 1.1561,
+      "step": 39200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.705165506420705e-05,
+      "loss": 1.2501,
+      "step": 39210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7050893414779045e-05,
+      "loss": 1.1018,
+      "step": 39220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.705013176535105e-05,
+      "loss": 1.2292,
+      "step": 39230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.704937011592305e-05,
+      "loss": 1.1618,
+      "step": 39240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7048608466495045e-05,
+      "loss": 1.1307,
+      "step": 39250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.704784681706704e-05,
+      "loss": 1.1372,
+      "step": 39260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7047085167639046e-05,
+      "loss": 1.0835,
+      "step": 39270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.704632351821104e-05,
+      "loss": 1.0961,
+      "step": 39280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.704556186878304e-05,
+      "loss": 1.0945,
+      "step": 39290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.704480021935504e-05,
+      "loss": 1.2928,
+      "step": 39300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7044038569927036e-05,
+      "loss": 1.3199,
+      "step": 39310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7043276920499034e-05,
+      "loss": 1.0173,
+      "step": 39320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.704251527107104e-05,
+      "loss": 1.1278,
+      "step": 39330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.704175362164303e-05,
+      "loss": 1.2076,
+      "step": 39340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.704099197221503e-05,
+      "loss": 1.1679,
+      "step": 39350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.704023032278703e-05,
+      "loss": 1.1436,
+      "step": 39360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703946867335903e-05,
+      "loss": 1.1503,
+      "step": 39370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7038707023931025e-05,
+      "loss": 1.1599,
+      "step": 39380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703794537450303e-05,
+      "loss": 1.2237,
+      "step": 39390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703718372507502e-05,
+      "loss": 1.0851,
+      "step": 39400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7036422075647024e-05,
+      "loss": 1.1845,
+      "step": 39410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703566042621902e-05,
+      "loss": 1.2777,
+      "step": 39420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703489877679102e-05,
+      "loss": 1.2583,
+      "step": 39430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7034137127363024e-05,
+      "loss": 1.0957,
+      "step": 39440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703337547793502e-05,
+      "loss": 1.1731,
+      "step": 39450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703261382850702e-05,
+      "loss": 1.1424,
+      "step": 39460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7031852179079016e-05,
+      "loss": 1.0662,
+      "step": 39470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7031090529651013e-05,
+      "loss": 1.2237,
+      "step": 39480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.703032888022301e-05,
+      "loss": 1.23,
+      "step": 39490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7029567230795015e-05,
+      "loss": 1.1003,
+      "step": 39500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7028805581367006e-05,
+      "loss": 1.2513,
+      "step": 39510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.702804393193901e-05,
+      "loss": 1.1646,
+      "step": 39520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.702728228251101e-05,
+      "loss": 1.1846,
+      "step": 39530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7026520633083005e-05,
+      "loss": 0.9566,
+      "step": 39540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7025758983655e-05,
+      "loss": 1.3261,
+      "step": 39550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.702499733422701e-05,
+      "loss": 1.163,
+      "step": 39560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7024235684799e-05,
+      "loss": 1.0939,
+      "step": 39570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7023474035371e-05,
+      "loss": 1.1427,
+      "step": 39580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7022712385943e-05,
+      "loss": 1.0677,
+      "step": 39590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7021950736515e-05,
+      "loss": 1.097,
+      "step": 39600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7021189087087e-05,
+      "loss": 1.2583,
+      "step": 39610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7020427437659e-05,
+      "loss": 1.071,
+      "step": 39620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7019665788230996e-05,
+      "loss": 1.235,
+      "step": 39630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701890413880299e-05,
+      "loss": 1.1618,
+      "step": 39640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7018142489375e-05,
+      "loss": 1.4607,
+      "step": 39650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701738083994699e-05,
+      "loss": 1.1497,
+      "step": 39660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701661919051899e-05,
+      "loss": 1.1707,
+      "step": 39670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701585754109099e-05,
+      "loss": 1.1954,
+      "step": 39680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701509589166299e-05,
+      "loss": 1.1087,
+      "step": 39690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7014334242234985e-05,
+      "loss": 1.2051,
+      "step": 39700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701357259280699e-05,
+      "loss": 1.2448,
+      "step": 39710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701281094337898e-05,
+      "loss": 1.1516,
+      "step": 39720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7012049293950984e-05,
+      "loss": 1.1061,
+      "step": 39730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.701128764452298e-05,
+      "loss": 1.1581,
+      "step": 39740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.701052599509498e-05,
+      "loss": 1.1131,
+      "step": 39750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.7009764345666976e-05,
+      "loss": 1.1287,
+      "step": 39760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700900269623898e-05,
+      "loss": 1.1004,
+      "step": 39770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700824104681097e-05,
+      "loss": 1.0647,
+      "step": 39780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.7007479397382976e-05,
+      "loss": 1.1761,
+      "step": 39790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700671774795497e-05,
+      "loss": 1.123,
+      "step": 39800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700595609852697e-05,
+      "loss": 1.1038,
+      "step": 39810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.7005194449098975e-05,
+      "loss": 1.1349,
+      "step": 39820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700443279967097e-05,
+      "loss": 1.1467,
+      "step": 39830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700367115024297e-05,
+      "loss": 1.2331,
+      "step": 39840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700290950081497e-05,
+      "loss": 0.9713,
+      "step": 39850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700214785138697e-05,
+      "loss": 1.1924,
+      "step": 39860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.700138620195896e-05,
+      "loss": 1.159,
+      "step": 39870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.7000624552530966e-05,
+      "loss": 1.1948,
+      "step": 39880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6999862903102964e-05,
+      "loss": 1.2104,
+      "step": 39890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.699910125367496e-05,
+      "loss": 1.1872,
+      "step": 39900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.699833960424696e-05,
+      "loss": 1.147,
+      "step": 39910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6997577954818956e-05,
+      "loss": 1.2275,
+      "step": 39920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6996816305390954e-05,
+      "loss": 1.1103,
+      "step": 39930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.699605465596296e-05,
+      "loss": 1.1116,
+      "step": 39940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.699529300653495e-05,
+      "loss": 1.2079,
+      "step": 39950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.699453135710695e-05,
+      "loss": 1.187,
+      "step": 39960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.699376970767895e-05,
+      "loss": 1.1663,
+      "step": 39970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.699300805825095e-05,
+      "loss": 1.1694,
+      "step": 39980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6992246408822945e-05,
+      "loss": 1.1801,
+      "step": 39990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.699148475939495e-05,
+      "loss": 1.2146,
+      "step": 40000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.699072310996695e-05,
+      "loss": 1.2069,
+      "step": 40010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6989961460538944e-05,
+      "loss": 1.1117,
+      "step": 40020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698919981111095e-05,
+      "loss": 1.2611,
+      "step": 40030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698843816168294e-05,
+      "loss": 1.1205,
+      "step": 40040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6987676512254944e-05,
+      "loss": 1.2071,
+      "step": 40050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698691486282694e-05,
+      "loss": 1.1263,
+      "step": 40060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698615321339894e-05,
+      "loss": 1.1713,
+      "step": 40070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6985391563970936e-05,
+      "loss": 1.0904,
+      "step": 40080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698462991454294e-05,
+      "loss": 1.1625,
+      "step": 40090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698386826511493e-05,
+      "loss": 1.197,
+      "step": 40100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6983106615686935e-05,
+      "loss": 1.223,
+      "step": 40110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698234496625893e-05,
+      "loss": 1.2608,
+      "step": 40120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698158331683093e-05,
+      "loss": 1.1618,
+      "step": 40130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698082166740293e-05,
+      "loss": 1.1646,
+      "step": 40140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.698006001797493e-05,
+      "loss": 1.0828,
+      "step": 40150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.697929836854692e-05,
+      "loss": 1.0773,
+      "step": 40160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.697853671911893e-05,
+      "loss": 1.2661,
+      "step": 40170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6977775069690924e-05,
+      "loss": 1.1757,
+      "step": 40180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697701342026292e-05,
+      "loss": 1.3354,
+      "step": 40190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6976251770834926e-05,
+      "loss": 1.1086,
+      "step": 40200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6975490121406923e-05,
+      "loss": 1.2434,
+      "step": 40210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697472847197892e-05,
+      "loss": 1.1999,
+      "step": 40220
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697396682255092e-05,
+      "loss": 1.042,
+      "step": 40230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697320517312292e-05,
+      "loss": 1.1691,
+      "step": 40240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697244352369491e-05,
+      "loss": 1.0931,
+      "step": 40250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697168187426692e-05,
+      "loss": 1.1465,
+      "step": 40260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6970920224838915e-05,
+      "loss": 1.2217,
+      "step": 40270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.697015857541091e-05,
+      "loss": 1.1329,
+      "step": 40280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.696939692598291e-05,
+      "loss": 1.0586,
+      "step": 40290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6968635276554914e-05,
+      "loss": 1.0986,
+      "step": 40300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6967873627126905e-05,
+      "loss": 1.0249,
+      "step": 40310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.696711197769891e-05,
+      "loss": 1.2141,
+      "step": 40320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6966350328270907e-05,
+      "loss": 1.0121,
+      "step": 40330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6965588678842904e-05,
+      "loss": 1.0464,
+      "step": 40340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.69648270294149e-05,
+      "loss": 1.2192,
+      "step": 40350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6964065379986906e-05,
+      "loss": 1.1901,
+      "step": 40360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6963303730558896e-05,
+      "loss": 1.2935,
+      "step": 40370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.69625420811309e-05,
+      "loss": 1.0248,
+      "step": 40380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.69617804317029e-05,
+      "loss": 1.2297,
+      "step": 40390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6961018782274896e-05,
+      "loss": 1.1358,
+      "step": 40400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.69602571328469e-05,
+      "loss": 1.2166,
+      "step": 40410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695949548341889e-05,
+      "loss": 1.1967,
+      "step": 40420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6958733833990895e-05,
+      "loss": 1.144,
+      "step": 40430
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695797218456289e-05,
+      "loss": 1.2357,
+      "step": 40440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695721053513489e-05,
+      "loss": 1.1774,
+      "step": 40450
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695644888570689e-05,
+      "loss": 1.325,
+      "step": 40460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695568723627889e-05,
+      "loss": 1.1793,
+      "step": 40470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695492558685088e-05,
+      "loss": 1.1376,
+      "step": 40480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6954163937422886e-05,
+      "loss": 1.109,
+      "step": 40490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6953402287994884e-05,
+      "loss": 1.2036,
+      "step": 40500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695264063856688e-05,
+      "loss": 1.1638,
+      "step": 40510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695187898913888e-05,
+      "loss": 1.4111,
+      "step": 40520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.695111733971088e-05,
+      "loss": 1.1783,
+      "step": 40530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6950355690282874e-05,
+      "loss": 1.048,
+      "step": 40540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.694959404085488e-05,
+      "loss": 1.1519,
+      "step": 40550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6948832391426875e-05,
+      "loss": 1.1579,
+      "step": 40560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.694807074199887e-05,
+      "loss": 1.1696,
+      "step": 40570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.694730909257087e-05,
+      "loss": 1.2533,
+      "step": 40580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6946547443142875e-05,
+      "loss": 1.2072,
+      "step": 40590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.694578579371487e-05,
+      "loss": 1.2067,
+      "step": 40600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.694502414428687e-05,
+      "loss": 1.1102,
+      "step": 40610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6944262494858874e-05,
+      "loss": 1.2546,
+      "step": 40620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6943500845430864e-05,
+      "loss": 1.2773,
+      "step": 40630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.694273919600287e-05,
+      "loss": 1.159,
+      "step": 40640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6941977546574866e-05,
+      "loss": 1.3566,
+      "step": 40650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6941215897146864e-05,
+      "loss": 1.2816,
+      "step": 40660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.694045424771886e-05,
+      "loss": 1.06,
+      "step": 40670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6939692598290865e-05,
+      "loss": 1.0668,
+      "step": 40680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6938930948862856e-05,
+      "loss": 1.2895,
+      "step": 40690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693816929943486e-05,
+      "loss": 1.1711,
+      "step": 40700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693740765000686e-05,
+      "loss": 1.1581,
+      "step": 40710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6936646000578855e-05,
+      "loss": 1.133,
+      "step": 40720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693588435115085e-05,
+      "loss": 1.1411,
+      "step": 40730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693512270172286e-05,
+      "loss": 1.2338,
+      "step": 40740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693436105229485e-05,
+      "loss": 1.1239,
+      "step": 40750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693359940286685e-05,
+      "loss": 0.9609,
+      "step": 40760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693283775343885e-05,
+      "loss": 1.0609,
+      "step": 40770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693207610401085e-05,
+      "loss": 1.0224,
+      "step": 40780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6931314454582844e-05,
+      "loss": 1.1587,
+      "step": 40790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.693055280515485e-05,
+      "loss": 1.136,
+      "step": 40800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6929791155726846e-05,
+      "loss": 1.208,
+      "step": 40810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6929029506298843e-05,
+      "loss": 1.1274,
+      "step": 40820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.692826785687085e-05,
+      "loss": 1.2347,
+      "step": 40830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.692750620744284e-05,
+      "loss": 1.0978,
+      "step": 40840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.692674455801484e-05,
+      "loss": 1.114,
+      "step": 40850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.692598290858683e-05,
+      "loss": 1.3098,
+      "step": 40860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.692522125915884e-05,
+      "loss": 1.0982,
+      "step": 40870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6924459609730835e-05,
+      "loss": 1.2059,
+      "step": 40880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.692369796030283e-05,
+      "loss": 1.1105,
+      "step": 40890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.692293631087483e-05,
+      "loss": 1.1394,
+      "step": 40900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6922174661446834e-05,
+      "loss": 1.1113,
+      "step": 40910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6921413012018825e-05,
+      "loss": 1.1067,
+      "step": 40920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.692065136259083e-05,
+      "loss": 1.2368,
+      "step": 40930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6919889713162827e-05,
+      "loss": 1.0903,
+      "step": 40940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6919128063734824e-05,
+      "loss": 1.1239,
+      "step": 40950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.691836641430682e-05,
+      "loss": 1.1252,
+      "step": 40960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6917604764878826e-05,
+      "loss": 1.1158,
+      "step": 40970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.691684311545082e-05,
+      "loss": 1.1552,
+      "step": 40980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.691608146602282e-05,
+      "loss": 1.1621,
+      "step": 40990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6915319816594825e-05,
+      "loss": 1.1703,
+      "step": 41000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6914558167166816e-05,
+      "loss": 1.1797,
+      "step": 41010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.691379651773882e-05,
+      "loss": 1.1826,
+      "step": 41020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.691303486831082e-05,
+      "loss": 1.0855,
+      "step": 41030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6912273218882815e-05,
+      "loss": 1.07,
+      "step": 41040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.691151156945481e-05,
+      "loss": 1.1516,
+      "step": 41050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6910749920026817e-05,
+      "loss": 1.1897,
+      "step": 41060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.690998827059881e-05,
+      "loss": 1.1672,
+      "step": 41070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.690922662117081e-05,
+      "loss": 1.1292,
+      "step": 41080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.690846497174281e-05,
+      "loss": 1.0864,
+      "step": 41090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6907703322314806e-05,
+      "loss": 1.0601,
+      "step": 41100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6906941672886804e-05,
+      "loss": 1.2201,
+      "step": 41110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.690618002345881e-05,
+      "loss": 1.2144,
+      "step": 41120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.69054183740308e-05,
+      "loss": 1.1599,
+      "step": 41130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.69046567246028e-05,
+      "loss": 1.0163,
+      "step": 41140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.69038950751748e-05,
+      "loss": 1.2007,
+      "step": 41150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.69031334257468e-05,
+      "loss": 1.0811,
+      "step": 41160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6902371776318795e-05,
+      "loss": 1.2323,
+      "step": 41170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.69016101268908e-05,
+      "loss": 1.2294,
+      "step": 41180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.69008484774628e-05,
+      "loss": 1.1518,
+      "step": 41190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6900086828034795e-05,
+      "loss": 1.1728,
+      "step": 41200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.68993251786068e-05,
+      "loss": 1.065,
+      "step": 41210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.689856352917879e-05,
+      "loss": 1.2097,
+      "step": 41220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6897801879750794e-05,
+      "loss": 1.2509,
+      "step": 41230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.689704023032279e-05,
+      "loss": 1.2061,
+      "step": 41240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.689627858089479e-05,
+      "loss": 1.1523,
+      "step": 41250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6895516931466786e-05,
+      "loss": 1.1281,
+      "step": 41260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.689475528203879e-05,
+      "loss": 1.0896,
+      "step": 41270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.689399363261078e-05,
+      "loss": 1.0662,
+      "step": 41280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6893231983182785e-05,
+      "loss": 1.017,
+      "step": 41290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6892470333754776e-05,
+      "loss": 1.1826,
+      "step": 41300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.689170868432678e-05,
+      "loss": 1.191,
+      "step": 41310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.689094703489878e-05,
+      "loss": 1.2417,
+      "step": 41320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6890185385470775e-05,
+      "loss": 1.174,
+      "step": 41330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688942373604277e-05,
+      "loss": 1.1637,
+      "step": 41340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688866208661478e-05,
+      "loss": 1.0835,
+      "step": 41350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6887900437186774e-05,
+      "loss": 1.0531,
+      "step": 41360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688713878775877e-05,
+      "loss": 1.3351,
+      "step": 41370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688637713833077e-05,
+      "loss": 1.1015,
+      "step": 41380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688561548890277e-05,
+      "loss": 1.0512,
+      "step": 41390
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688485383947477e-05,
+      "loss": 1.0246,
+      "step": 41400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688409219004677e-05,
+      "loss": 1.3625,
+      "step": 41410
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6883330540618766e-05,
+      "loss": 1.2751,
+      "step": 41420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6882568891190763e-05,
+      "loss": 1.1149,
+      "step": 41430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688180724176277e-05,
+      "loss": 1.3183,
+      "step": 41440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688104559233476e-05,
+      "loss": 1.069,
+      "step": 41450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.688028394290676e-05,
+      "loss": 1.1808,
+      "step": 41460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.687952229347876e-05,
+      "loss": 1.2616,
+      "step": 41470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.687876064405076e-05,
+      "loss": 1.2402,
+      "step": 41480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6877998994622755e-05,
+      "loss": 1.0729,
+      "step": 41490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.687723734519476e-05,
+      "loss": 1.0991,
+      "step": 41500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.687647569576675e-05,
+      "loss": 1.2512,
+      "step": 41510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6875714046338754e-05,
+      "loss": 1.1489,
+      "step": 41520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.687495239691075e-05,
+      "loss": 1.2117,
+      "step": 41530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.687419074748275e-05,
+      "loss": 0.9551,
+      "step": 41540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6873429098054747e-05,
+      "loss": 1.1002,
+      "step": 41550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.687266744862675e-05,
+      "loss": 1.2589,
+      "step": 41560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.687190579919875e-05,
+      "loss": 1.0657,
+      "step": 41570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6871144149770746e-05,
+      "loss": 1.2029,
+      "step": 41580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.687038250034274e-05,
+      "loss": 1.1266,
+      "step": 41590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686962085091474e-05,
+      "loss": 1.1986,
+      "step": 41600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6868859201486745e-05,
+      "loss": 1.1575,
+      "step": 41610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686809755205874e-05,
+      "loss": 1.1438,
+      "step": 41620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686733590263074e-05,
+      "loss": 1.1879,
+      "step": 41630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686657425320274e-05,
+      "loss": 1.1348,
+      "step": 41640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686581260377474e-05,
+      "loss": 1.0866,
+      "step": 41650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686505095434673e-05,
+      "loss": 1.127,
+      "step": 41660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6864289304918737e-05,
+      "loss": 1.0334,
+      "step": 41670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6863527655490734e-05,
+      "loss": 1.2643,
+      "step": 41680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686276600606273e-05,
+      "loss": 1.0313,
+      "step": 41690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686200435663473e-05,
+      "loss": 1.2998,
+      "step": 41700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.686124270720673e-05,
+      "loss": 1.0205,
+      "step": 41710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6860481057778724e-05,
+      "loss": 1.0845,
+      "step": 41720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685971940835073e-05,
+      "loss": 1.1865,
+      "step": 41730
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6858957758922726e-05,
+      "loss": 1.1555,
+      "step": 41740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685819610949472e-05,
+      "loss": 1.0714,
+      "step": 41750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685743446006672e-05,
+      "loss": 1.1247,
+      "step": 41760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685667281063872e-05,
+      "loss": 1.0805,
+      "step": 41770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685591116121072e-05,
+      "loss": 1.1014,
+      "step": 41780
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685514951178272e-05,
+      "loss": 1.2332,
+      "step": 41790
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685438786235472e-05,
+      "loss": 1.0963,
+      "step": 41800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6853626212926715e-05,
+      "loss": 1.1287,
+      "step": 41810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685286456349872e-05,
+      "loss": 1.1825,
+      "step": 41820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685210291407071e-05,
+      "loss": 1.1504,
+      "step": 41830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6851341264642714e-05,
+      "loss": 1.1779,
+      "step": 41840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.685057961521471e-05,
+      "loss": 1.0675,
+      "step": 41850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.684981796578671e-05,
+      "loss": 1.2608,
+      "step": 41860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6849056316358706e-05,
+      "loss": 1.1505,
+      "step": 41870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.684829466693071e-05,
+      "loss": 1.2436,
+      "step": 41880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.68475330175027e-05,
+      "loss": 1.1323,
+      "step": 41890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6846771368074705e-05,
+      "loss": 1.1673,
+      "step": 41900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.68460097186467e-05,
+      "loss": 1.2435,
+      "step": 41910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.68452480692187e-05,
+      "loss": 1.1671,
+      "step": 41920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.68444864197907e-05,
+      "loss": 1.1268,
+      "step": 41930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.68437247703627e-05,
+      "loss": 1.0961,
+      "step": 41940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.68429631209347e-05,
+      "loss": 1.247,
+      "step": 41950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.68422014715067e-05,
+      "loss": 1.1144,
+      "step": 41960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6841439822078694e-05,
+      "loss": 1.1145,
+      "step": 41970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.684067817265069e-05,
+      "loss": 1.1019,
+      "step": 41980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6839916523222696e-05,
+      "loss": 1.1453,
+      "step": 41990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6839154873794694e-05,
+      "loss": 1.2206,
+      "step": 42000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683839322436669e-05,
+      "loss": 1.2161,
+      "step": 42010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683763157493869e-05,
+      "loss": 1.157,
+      "step": 42020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683686992551069e-05,
+      "loss": 1.2144,
+      "step": 42030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6836108276082683e-05,
+      "loss": 1.1211,
+      "step": 42040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683534662665469e-05,
+      "loss": 1.1972,
+      "step": 42050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6834584977226685e-05,
+      "loss": 1.212,
+      "step": 42060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683382332779868e-05,
+      "loss": 1.1418,
+      "step": 42070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683306167837068e-05,
+      "loss": 1.1599,
+      "step": 42080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6832300028942684e-05,
+      "loss": 1.0554,
+      "step": 42090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6831538379514675e-05,
+      "loss": 1.1177,
+      "step": 42100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683077673008668e-05,
+      "loss": 1.086,
+      "step": 42110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683001508065868e-05,
+      "loss": 1.0377,
+      "step": 42120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6829253431230674e-05,
+      "loss": 1.2387,
+      "step": 42130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.682849178180267e-05,
+      "loss": 1.1586,
+      "step": 42140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6827730132374676e-05,
+      "loss": 1.1389,
+      "step": 42150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682696848294667e-05,
+      "loss": 1.224,
+      "step": 42160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682620683351867e-05,
+      "loss": 1.0966,
+      "step": 42170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682544518409067e-05,
+      "loss": 1.2563,
+      "step": 42180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6824683534662666e-05,
+      "loss": 1.1453,
+      "step": 42190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682392188523467e-05,
+      "loss": 1.0538,
+      "step": 42200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682316023580667e-05,
+      "loss": 1.0842,
+      "step": 42210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6822398586378665e-05,
+      "loss": 1.2769,
+      "step": 42220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682163693695066e-05,
+      "loss": 1.1795,
+      "step": 42230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682087528752266e-05,
+      "loss": 1.234,
+      "step": 42240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.682011363809466e-05,
+      "loss": 1.112,
+      "step": 42250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681935198866666e-05,
+      "loss": 1.0763,
+      "step": 42260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681859033923865e-05,
+      "loss": 1.266,
+      "step": 42270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6817828689810657e-05,
+      "loss": 1.164,
+      "step": 42280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6817067040382654e-05,
+      "loss": 1.1499,
+      "step": 42290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681630539095465e-05,
+      "loss": 1.0945,
+      "step": 42300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681554374152665e-05,
+      "loss": 1.2244,
+      "step": 42310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681478209209865e-05,
+      "loss": 1.0913,
+      "step": 42320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681402044267065e-05,
+      "loss": 1.1034,
+      "step": 42330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681325879324265e-05,
+      "loss": 1.2516,
+      "step": 42340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6812497143814646e-05,
+      "loss": 1.1865,
+      "step": 42350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681173549438664e-05,
+      "loss": 1.0527,
+      "step": 42360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.681097384495865e-05,
+      "loss": 1.1801,
+      "step": 42370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6810212195530645e-05,
+      "loss": 1.2276,
+      "step": 42380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.680945054610264e-05,
+      "loss": 1.1026,
+      "step": 42390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.680868889667464e-05,
+      "loss": 1.1167,
+      "step": 42400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6807927247246644e-05,
+      "loss": 1.0755,
+      "step": 42410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6807165597818635e-05,
+      "loss": 1.2714,
+      "step": 42420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.680640394839064e-05,
+      "loss": 1.1713,
+      "step": 42430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6805642298962636e-05,
+      "loss": 1.1549,
+      "step": 42440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6804880649534634e-05,
+      "loss": 1.0493,
+      "step": 42450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.680411900010663e-05,
+      "loss": 1.2123,
+      "step": 42460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6803357350678636e-05,
+      "loss": 1.2009,
+      "step": 42470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6802595701250626e-05,
+      "loss": 1.0944,
+      "step": 42480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.680183405182263e-05,
+      "loss": 1.0689,
+      "step": 42490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.680107240239463e-05,
+      "loss": 1.152,
+      "step": 42500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6800310752966625e-05,
+      "loss": 1.1169,
+      "step": 42510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679954910353862e-05,
+      "loss": 1.2198,
+      "step": 42520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679878745411063e-05,
+      "loss": 1.1507,
+      "step": 42530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6798025804682625e-05,
+      "loss": 1.3663,
+      "step": 42540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679726415525462e-05,
+      "loss": 1.2631,
+      "step": 42550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679650250582662e-05,
+      "loss": 1.1914,
+      "step": 42560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679574085639862e-05,
+      "loss": 1.3324,
+      "step": 42570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679497920697062e-05,
+      "loss": 1.1408,
+      "step": 42580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.679421755754262e-05,
+      "loss": 1.0199,
+      "step": 42590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6793455908114616e-05,
+      "loss": 1.0872,
+      "step": 42600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6792694258686614e-05,
+      "loss": 1.1115,
+      "step": 42610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.679193260925862e-05,
+      "loss": 1.1408,
+      "step": 42620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.679117095983061e-05,
+      "loss": 1.2197,
+      "step": 42630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.679040931040261e-05,
+      "loss": 1.0708,
+      "step": 42640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.678964766097461e-05,
+      "loss": 0.9909,
+      "step": 42650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.678888601154661e-05,
+      "loss": 1.1814,
+      "step": 42660
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6788124362118605e-05,
+      "loss": 1.0984,
+      "step": 42670
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.678736271269061e-05,
+      "loss": 0.9688,
+      "step": 42680
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67866010632626e-05,
+      "loss": 1.1743,
+      "step": 42690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6785839413834604e-05,
+      "loss": 1.2059,
+      "step": 42700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67850777644066e-05,
+      "loss": 1.0727,
+      "step": 42710
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67843161149786e-05,
+      "loss": 1.2991,
+      "step": 42720
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67835544655506e-05,
+      "loss": 1.2565,
+      "step": 42730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6782792816122594e-05,
+      "loss": 1.0713,
+      "step": 42740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67820311666946e-05,
+      "loss": 1.1535,
+      "step": 42750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6781269517266596e-05,
+      "loss": 1.0464,
+      "step": 42760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6780507867838593e-05,
+      "loss": 1.0321,
+      "step": 42770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.677974621841059e-05,
+      "loss": 1.3021,
+      "step": 42780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6778984568982595e-05,
+      "loss": 1.0457,
+      "step": 42790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6778222919554586e-05,
+      "loss": 1.2014,
+      "step": 42800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.677746127012659e-05,
+      "loss": 1.1556,
+      "step": 42810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.677669962069859e-05,
+      "loss": 1.2333,
+      "step": 42820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6775937971270585e-05,
+      "loss": 1.0922,
+      "step": 42830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.677517632184258e-05,
+      "loss": 1.2242,
+      "step": 42840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.677441467241459e-05,
+      "loss": 1.1139,
+      "step": 42850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.677365302298658e-05,
+      "loss": 1.164,
+      "step": 42860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.677289137355858e-05,
+      "loss": 1.2673,
+      "step": 42870
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.677212972413058e-05,
+      "loss": 1.1815,
+      "step": 42880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6771368074702577e-05,
+      "loss": 1.2464,
+      "step": 42890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6770606425274574e-05,
+      "loss": 1.0647,
+      "step": 42900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676984477584658e-05,
+      "loss": 1.1058,
+      "step": 42910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6769083126418576e-05,
+      "loss": 1.1687,
+      "step": 42920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676832147699057e-05,
+      "loss": 1.1876,
+      "step": 42930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676755982756257e-05,
+      "loss": 1.2212,
+      "step": 42940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676679817813457e-05,
+      "loss": 1.1532,
+      "step": 42950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676603652870657e-05,
+      "loss": 1.116,
+      "step": 42960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676527487927857e-05,
+      "loss": 1.1677,
+      "step": 42970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676451322985057e-05,
+      "loss": 1.178,
+      "step": 42980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6763751580422565e-05,
+      "loss": 1.1432,
+      "step": 42990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676298993099457e-05,
+      "loss": 1.1497,
+      "step": 43000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676222828156656e-05,
+      "loss": 1.1225,
+      "step": 43010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6761466632138564e-05,
+      "loss": 1.2255,
+      "step": 43020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.676070498271056e-05,
+      "loss": 1.136,
+      "step": 43030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675994333328256e-05,
+      "loss": 1.0212,
+      "step": 43040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6759181683854556e-05,
+      "loss": 1.0949,
+      "step": 43050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675842003442656e-05,
+      "loss": 1.1474,
+      "step": 43060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675765838499855e-05,
+      "loss": 1.1165,
+      "step": 43070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6756896735570556e-05,
+      "loss": 1.1789,
+      "step": 43080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675613508614255e-05,
+      "loss": 1.1106,
+      "step": 43090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675537343671455e-05,
+      "loss": 1.1646,
+      "step": 43100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675461178728655e-05,
+      "loss": 1.0839,
+      "step": 43110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675385013785855e-05,
+      "loss": 1.343,
+      "step": 43120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675308848843055e-05,
+      "loss": 1.2317,
+      "step": 43130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675232683900255e-05,
+      "loss": 1.0369,
+      "step": 43140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6751565189574545e-05,
+      "loss": 1.2451,
+      "step": 43150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.675080354014654e-05,
+      "loss": 1.1534,
+      "step": 43160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6750041890718546e-05,
+      "loss": 1.2522,
+      "step": 43170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.674928024129054e-05,
+      "loss": 1.1599,
+      "step": 43180
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.674851859186254e-05,
+      "loss": 1.1314,
+      "step": 43190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.674775694243454e-05,
+      "loss": 1.2522,
+      "step": 43200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6746995293006536e-05,
+      "loss": 1.1438,
+      "step": 43210
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6746233643578534e-05,
+      "loss": 1.2727,
+      "step": 43220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.674547199415054e-05,
+      "loss": 1.1344,
+      "step": 43230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.674471034472253e-05,
+      "loss": 1.1345,
+      "step": 43240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.674394869529453e-05,
+      "loss": 1.2036,
+      "step": 43250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.674318704586653e-05,
+      "loss": 1.0017,
+      "step": 43260
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.674242539643853e-05,
+      "loss": 1.2141,
+      "step": 43270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6741663747010525e-05,
+      "loss": 1.1364,
+      "step": 43280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.674090209758253e-05,
+      "loss": 1.0947,
+      "step": 43290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.674014044815453e-05,
+      "loss": 1.1082,
+      "step": 43300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6739378798726524e-05,
+      "loss": 1.1995,
+      "step": 43310
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673861714929852e-05,
+      "loss": 1.2961,
+      "step": 43320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673785549987052e-05,
+      "loss": 1.3215,
+      "step": 43330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6737093850442524e-05,
+      "loss": 1.0738,
+      "step": 43340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673633220101452e-05,
+      "loss": 1.0973,
+      "step": 43350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673557055158652e-05,
+      "loss": 1.0546,
+      "step": 43360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6734808902158516e-05,
+      "loss": 1.0632,
+      "step": 43370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673404725273052e-05,
+      "loss": 1.0994,
+      "step": 43380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673328560330251e-05,
+      "loss": 1.1994,
+      "step": 43390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6732523953874515e-05,
+      "loss": 1.123,
+      "step": 43400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673176230444651e-05,
+      "loss": 1.1893,
+      "step": 43410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673100065501851e-05,
+      "loss": 1.1199,
+      "step": 43420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.673023900559051e-05,
+      "loss": 1.1938,
+      "step": 43430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.672947735616251e-05,
+      "loss": 1.2219,
+      "step": 43440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.67287157067345e-05,
+      "loss": 1.0736,
+      "step": 43450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.672795405730651e-05,
+      "loss": 1.0673,
+      "step": 43460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6727192407878504e-05,
+      "loss": 1.1735,
+      "step": 43470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.67264307584505e-05,
+      "loss": 1.243,
+      "step": 43480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.67256691090225e-05,
+      "loss": 1.0754,
+      "step": 43490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.67249074595945e-05,
+      "loss": 1.1252,
+      "step": 43500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.67241458101665e-05,
+      "loss": 1.2185,
+      "step": 43510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.67233841607385e-05,
+      "loss": 1.2262,
+      "step": 43520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6722622511310496e-05,
+      "loss": 1.1303,
+      "step": 43530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.672186086188249e-05,
+      "loss": 1.1517,
+      "step": 43540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.67210992124545e-05,
+      "loss": 1.1946,
+      "step": 43550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6720337563026495e-05,
+      "loss": 0.9925,
+      "step": 43560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671957591359849e-05,
+      "loss": 1.0298,
+      "step": 43570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671881426417049e-05,
+      "loss": 1.134,
+      "step": 43580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6718052614742494e-05,
+      "loss": 1.2751,
+      "step": 43590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6717290965314485e-05,
+      "loss": 1.2532,
+      "step": 43600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671652931588649e-05,
+      "loss": 1.2217,
+      "step": 43610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671576766645848e-05,
+      "loss": 1.1386,
+      "step": 43620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6715006017030484e-05,
+      "loss": 1.1705,
+      "step": 43630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671424436760248e-05,
+      "loss": 1.0623,
+      "step": 43640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671348271817448e-05,
+      "loss": 1.3309,
+      "step": 43650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6712721068746476e-05,
+      "loss": 1.0827,
+      "step": 43660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671195941931848e-05,
+      "loss": 1.3382,
+      "step": 43670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.671119776989047e-05,
+      "loss": 1.0835,
+      "step": 43680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6710436120462476e-05,
+      "loss": 1.2375,
+      "step": 43690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670967447103447e-05,
+      "loss": 1.1341,
+      "step": 43700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670891282160647e-05,
+      "loss": 1.1733,
+      "step": 43710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6708151172178475e-05,
+      "loss": 1.1166,
+      "step": 43720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670738952275047e-05,
+      "loss": 1.3078,
+      "step": 43730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670662787332247e-05,
+      "loss": 0.9699,
+      "step": 43740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670586622389447e-05,
+      "loss": 1.1747,
+      "step": 43750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670510457446647e-05,
+      "loss": 1.148,
+      "step": 43760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670434292503846e-05,
+      "loss": 1.1429,
+      "step": 43770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6703581275610466e-05,
+      "loss": 1.0551,
+      "step": 43780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6702819626182464e-05,
+      "loss": 1.1607,
+      "step": 43790
+    },
+    {
+      "epoch": 2.0,
+      "eval_cer": 0.6298998809124814,
+      "eval_em": 0.007379375591296121,
+      "eval_f1": 0.007379375591296121,
+      "eval_loss": 0.9932093620300293,
+      "eval_runtime": 1008.9215,
+      "eval_samples_per_second": 10.477,
+      "eval_steps_per_second": 1.31,
+      "eval_wer": 0.9926206244087039,
+      "step": 43799
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670205797675446e-05,
+      "loss": 1.1399,
+      "step": 43800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670129632732646e-05,
+      "loss": 1.2411,
+      "step": 43810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.670053467789846e-05,
+      "loss": 1.2053,
+      "step": 43820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6699773028470454e-05,
+      "loss": 1.026,
+      "step": 43830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.669901137904246e-05,
+      "loss": 1.1144,
+      "step": 43840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6698249729614455e-05,
+      "loss": 1.1313,
+      "step": 43850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.669748808018645e-05,
+      "loss": 1.0478,
+      "step": 43860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.669672643075845e-05,
+      "loss": 1.1006,
+      "step": 43870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6695964781330455e-05,
+      "loss": 0.9967,
+      "step": 43880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6695203131902445e-05,
+      "loss": 1.1105,
+      "step": 43890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.669444148247445e-05,
+      "loss": 1.0018,
+      "step": 43900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.669367983304645e-05,
+      "loss": 1.2584,
+      "step": 43910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6692918183618444e-05,
+      "loss": 1.094,
+      "step": 43920
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.669215653419045e-05,
+      "loss": 1.1237,
+      "step": 43930
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6691394884762446e-05,
+      "loss": 1.1505,
+      "step": 43940
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6690633235334444e-05,
+      "loss": 1.136,
+      "step": 43950
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668987158590644e-05,
+      "loss": 1.1481,
+      "step": 43960
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6689109936478445e-05,
+      "loss": 1.1229,
+      "step": 43970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6688348287050436e-05,
+      "loss": 1.0906,
+      "step": 43980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668758663762244e-05,
+      "loss": 1.1452,
+      "step": 43990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668682498819444e-05,
+      "loss": 1.2149,
+      "step": 44000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6686063338766435e-05,
+      "loss": 1.1267,
+      "step": 44010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668530168933843e-05,
+      "loss": 1.1627,
+      "step": 44020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668454003991044e-05,
+      "loss": 0.924,
+      "step": 44030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668377839048243e-05,
+      "loss": 1.1608,
+      "step": 44040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668301674105443e-05,
+      "loss": 1.0371,
+      "step": 44050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668225509162643e-05,
+      "loss": 1.0394,
+      "step": 44060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.668149344219843e-05,
+      "loss": 1.1422,
+      "step": 44070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6680731792770424e-05,
+      "loss": 1.0693,
+      "step": 44080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.667997014334242e-05,
+      "loss": 1.0764,
+      "step": 44090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6679208493914426e-05,
+      "loss": 1.1249,
+      "step": 44100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.667844684448642e-05,
+      "loss": 1.2675,
+      "step": 44110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.667768519505842e-05,
+      "loss": 1.0688,
+      "step": 44120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667692354563042e-05,
+      "loss": 1.119,
+      "step": 44130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667616189620242e-05,
+      "loss": 1.0663,
+      "step": 44140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667540024677441e-05,
+      "loss": 1.2443,
+      "step": 44150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667463859734642e-05,
+      "loss": 1.1388,
+      "step": 44160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6673876947918415e-05,
+      "loss": 1.1128,
+      "step": 44170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667311529849041e-05,
+      "loss": 1.1168,
+      "step": 44180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667235364906241e-05,
+      "loss": 1.1244,
+      "step": 44190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6671591999634414e-05,
+      "loss": 1.2478,
+      "step": 44200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6670830350206405e-05,
+      "loss": 1.1403,
+      "step": 44210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.667006870077841e-05,
+      "loss": 1.1253,
+      "step": 44220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6669307051350407e-05,
+      "loss": 1.1484,
+      "step": 44230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6668545401922404e-05,
+      "loss": 1.0724,
+      "step": 44240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66677837524944e-05,
+      "loss": 1.0056,
+      "step": 44250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6667022103066406e-05,
+      "loss": 1.2036,
+      "step": 44260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6666260453638396e-05,
+      "loss": 1.1001,
+      "step": 44270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66654988042104e-05,
+      "loss": 1.0646,
+      "step": 44280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66647371547824e-05,
+      "loss": 1.2212,
+      "step": 44290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6663975505354396e-05,
+      "loss": 1.1855,
+      "step": 44300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66632138559264e-05,
+      "loss": 1.0274,
+      "step": 44310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66624522064984e-05,
+      "loss": 1.0829,
+      "step": 44320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6661690557070395e-05,
+      "loss": 0.9551,
+      "step": 44330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.666092890764239e-05,
+      "loss": 1.1509,
+      "step": 44340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6660167258214396e-05,
+      "loss": 1.0063,
+      "step": 44350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665940560878639e-05,
+      "loss": 1.1276,
+      "step": 44360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665864395935839e-05,
+      "loss": 1.2,
+      "step": 44370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665788230993039e-05,
+      "loss": 1.0531,
+      "step": 44380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6657120660502386e-05,
+      "loss": 1.0346,
+      "step": 44390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6656359011074384e-05,
+      "loss": 1.0623,
+      "step": 44400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665559736164639e-05,
+      "loss": 1.0955,
+      "step": 44410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665483571221838e-05,
+      "loss": 1.1276,
+      "step": 44420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665407406279038e-05,
+      "loss": 1.0893,
+      "step": 44430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665331241336238e-05,
+      "loss": 1.12,
+      "step": 44440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665255076393438e-05,
+      "loss": 1.1389,
+      "step": 44450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6651789114506375e-05,
+      "loss": 1.0588,
+      "step": 44460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665102746507838e-05,
+      "loss": 1.2442,
+      "step": 44470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665026581565037e-05,
+      "loss": 1.0374,
+      "step": 44480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6649504166222375e-05,
+      "loss": 1.1768,
+      "step": 44490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.664874251679437e-05,
+      "loss": 1.0953,
+      "step": 44500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.664798086736637e-05,
+      "loss": 1.1827,
+      "step": 44510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6647219217938374e-05,
+      "loss": 0.9965,
+      "step": 44520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6646457568510364e-05,
+      "loss": 1.146,
+      "step": 44530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.664569591908237e-05,
+      "loss": 1.0649,
+      "step": 44540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6644934269654366e-05,
+      "loss": 1.1356,
+      "step": 44550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6644172620226364e-05,
+      "loss": 1.0525,
+      "step": 44560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.664341097079836e-05,
+      "loss": 1.1323,
+      "step": 44570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6642649321370365e-05,
+      "loss": 1.1516,
+      "step": 44580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6641887671942356e-05,
+      "loss": 1.0833,
+      "step": 44590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.664112602251436e-05,
+      "loss": 1.1867,
+      "step": 44600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.664036437308636e-05,
+      "loss": 1.1973,
+      "step": 44610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6639602723658355e-05,
+      "loss": 1.1051,
+      "step": 44620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663884107423035e-05,
+      "loss": 1.0777,
+      "step": 44630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663807942480236e-05,
+      "loss": 1.0674,
+      "step": 44640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663731777537435e-05,
+      "loss": 1.0974,
+      "step": 44650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663655612594635e-05,
+      "loss": 1.0983,
+      "step": 44660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663579447651835e-05,
+      "loss": 0.9771,
+      "step": 44670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663503282709035e-05,
+      "loss": 1.2275,
+      "step": 44680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6634271177662344e-05,
+      "loss": 1.1196,
+      "step": 44690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663350952823435e-05,
+      "loss": 1.1418,
+      "step": 44700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6632747878806346e-05,
+      "loss": 1.0531,
+      "step": 44710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663198622937834e-05,
+      "loss": 1.0189,
+      "step": 44720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663122457995035e-05,
+      "loss": 0.9713,
+      "step": 44730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.663046293052234e-05,
+      "loss": 1.1264,
+      "step": 44740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.662970128109434e-05,
+      "loss": 1.0349,
+      "step": 44750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.662893963166634e-05,
+      "loss": 1.2507,
+      "step": 44760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.662817798223834e-05,
+      "loss": 1.2291,
+      "step": 44770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6627416332810335e-05,
+      "loss": 1.0525,
+      "step": 44780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662665468338234e-05,
+      "loss": 1.0451,
+      "step": 44790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662589303395433e-05,
+      "loss": 0.9773,
+      "step": 44800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6625131384526334e-05,
+      "loss": 1.0913,
+      "step": 44810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662436973509833e-05,
+      "loss": 1.1231,
+      "step": 44820
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662360808567033e-05,
+      "loss": 1.08,
+      "step": 44830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6622846436242327e-05,
+      "loss": 1.0146,
+      "step": 44840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662208478681433e-05,
+      "loss": 1.2435,
+      "step": 44850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662132313738632e-05,
+      "loss": 1.0721,
+      "step": 44860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6620561487958326e-05,
+      "loss": 1.1309,
+      "step": 44870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661979983853032e-05,
+      "loss": 1.0751,
+      "step": 44880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661903818910232e-05,
+      "loss": 1.0,
+      "step": 44890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6618276539674325e-05,
+      "loss": 1.1733,
+      "step": 44900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661751489024632e-05,
+      "loss": 1.1363,
+      "step": 44910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661675324081832e-05,
+      "loss": 1.0998,
+      "step": 44920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661599159139032e-05,
+      "loss": 1.1199,
+      "step": 44930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661522994196232e-05,
+      "loss": 1.1401,
+      "step": 44940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661446829253431e-05,
+      "loss": 1.1379,
+      "step": 44950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6613706643106316e-05,
+      "loss": 1.1371,
+      "step": 44960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6612944993678314e-05,
+      "loss": 1.081,
+      "step": 44970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661218334425031e-05,
+      "loss": 1.0639,
+      "step": 44980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.661142169482231e-05,
+      "loss": 0.9458,
+      "step": 44990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6610660045394306e-05,
+      "loss": 1.236,
+      "step": 45000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6609898395966304e-05,
+      "loss": 1.1914,
+      "step": 45010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.660913674653831e-05,
+      "loss": 1.1499,
+      "step": 45020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66083750971103e-05,
+      "loss": 0.9668,
+      "step": 45030
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66076134476823e-05,
+      "loss": 1.1192,
+      "step": 45040
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66068517982543e-05,
+      "loss": 0.9862,
+      "step": 45050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66060901488263e-05,
+      "loss": 1.2513,
+      "step": 45060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6605328499398295e-05,
+      "loss": 1.1632,
+      "step": 45070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66045668499703e-05,
+      "loss": 1.1271,
+      "step": 45080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66038052005423e-05,
+      "loss": 1.0362,
+      "step": 45090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6603043551114295e-05,
+      "loss": 1.143,
+      "step": 45100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.66022819016863e-05,
+      "loss": 1.1453,
+      "step": 45110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.660152025225829e-05,
+      "loss": 1.0512,
+      "step": 45120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6600758602830294e-05,
+      "loss": 1.1501,
+      "step": 45130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659999695340229e-05,
+      "loss": 1.0424,
+      "step": 45140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659923530397429e-05,
+      "loss": 1.0411,
+      "step": 45150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6598473654546286e-05,
+      "loss": 1.0473,
+      "step": 45160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659771200511829e-05,
+      "loss": 1.077,
+      "step": 45170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659695035569028e-05,
+      "loss": 1.1789,
+      "step": 45180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6596188706262285e-05,
+      "loss": 1.1604,
+      "step": 45190
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659542705683428e-05,
+      "loss": 1.1829,
+      "step": 45200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659466540740628e-05,
+      "loss": 1.1022,
+      "step": 45210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.659390375797828e-05,
+      "loss": 0.9974,
+      "step": 45220
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659314210855028e-05,
+      "loss": 1.2121,
+      "step": 45230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659238045912227e-05,
+      "loss": 1.0743,
+      "step": 45240
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659161880969428e-05,
+      "loss": 1.2742,
+      "step": 45250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6590857160266274e-05,
+      "loss": 1.0724,
+      "step": 45260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659009551083827e-05,
+      "loss": 1.2149,
+      "step": 45270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658933386141027e-05,
+      "loss": 1.1446,
+      "step": 45280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6588572211982274e-05,
+      "loss": 1.1168,
+      "step": 45290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658781056255427e-05,
+      "loss": 1.0873,
+      "step": 45300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658704891312627e-05,
+      "loss": 1.0366,
+      "step": 45310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658628726369827e-05,
+      "loss": 1.1094,
+      "step": 45320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658552561427026e-05,
+      "loss": 1.1778,
+      "step": 45330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658476396484227e-05,
+      "loss": 1.0045,
+      "step": 45340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6584002315414265e-05,
+      "loss": 1.2112,
+      "step": 45350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658324066598626e-05,
+      "loss": 1.0552,
+      "step": 45360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658247901655826e-05,
+      "loss": 1.0029,
+      "step": 45370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6581717367130264e-05,
+      "loss": 1.055,
+      "step": 45380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6580955717702255e-05,
+      "loss": 1.0662,
+      "step": 45390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658019406827426e-05,
+      "loss": 1.1659,
+      "step": 45400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.657943241884626e-05,
+      "loss": 1.1212,
+      "step": 45410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6578670769418254e-05,
+      "loss": 1.0662,
+      "step": 45420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.657790911999025e-05,
+      "loss": 1.1866,
+      "step": 45430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6577147470562256e-05,
+      "loss": 1.0828,
+      "step": 45440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6576385821134247e-05,
+      "loss": 1.1464,
+      "step": 45450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.657562417170625e-05,
+      "loss": 1.2319,
+      "step": 45460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.657486252227825e-05,
+      "loss": 1.1431,
+      "step": 45470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6574100872850246e-05,
+      "loss": 1.1662,
+      "step": 45480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.657333922342224e-05,
+      "loss": 1.0824,
+      "step": 45490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.657257757399424e-05,
+      "loss": 1.1315,
+      "step": 45500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6571815924566245e-05,
+      "loss": 1.1193,
+      "step": 45510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.657105427513824e-05,
+      "loss": 1.1248,
+      "step": 45520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.657029262571024e-05,
+      "loss": 1.1351,
+      "step": 45530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656953097628224e-05,
+      "loss": 1.126,
+      "step": 45540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656876932685424e-05,
+      "loss": 1.0548,
+      "step": 45550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656800767742623e-05,
+      "loss": 1.1231,
+      "step": 45560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6567246027998236e-05,
+      "loss": 1.1795,
+      "step": 45570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6566484378570234e-05,
+      "loss": 1.0381,
+      "step": 45580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656572272914223e-05,
+      "loss": 1.1209,
+      "step": 45590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656496107971423e-05,
+      "loss": 1.1186,
+      "step": 45600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656419943028623e-05,
+      "loss": 1.0587,
+      "step": 45610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6563437780858224e-05,
+      "loss": 1.0758,
+      "step": 45620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656267613143023e-05,
+      "loss": 1.1144,
+      "step": 45630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6561914482002225e-05,
+      "loss": 0.9929,
+      "step": 45640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656115283257422e-05,
+      "loss": 1.2311,
+      "step": 45650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.656039118314622e-05,
+      "loss": 0.962,
+      "step": 45660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6559629533718225e-05,
+      "loss": 1.0499,
+      "step": 45670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.655886788429022e-05,
+      "loss": 1.0337,
+      "step": 45680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.655810623486222e-05,
+      "loss": 1.1353,
+      "step": 45690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6557344585434224e-05,
+      "loss": 1.0279,
+      "step": 45700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6556582936006215e-05,
+      "loss": 1.106,
+      "step": 45710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.655582128657822e-05,
+      "loss": 1.0668,
+      "step": 45720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6555059637150216e-05,
+      "loss": 1.0097,
+      "step": 45730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6554297987722214e-05,
+      "loss": 1.1668,
+      "step": 45740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.655353633829421e-05,
+      "loss": 1.2509,
+      "step": 45750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6552774688866215e-05,
+      "loss": 1.1469,
+      "step": 45760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6552013039438206e-05,
+      "loss": 1.1596,
+      "step": 45770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.655125139001021e-05,
+      "loss": 1.1235,
+      "step": 45780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.655048974058221e-05,
+      "loss": 1.2319,
+      "step": 45790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6549728091154205e-05,
+      "loss": 1.0361,
+      "step": 45800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65489664417262e-05,
+      "loss": 0.9687,
+      "step": 45810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.654820479229821e-05,
+      "loss": 1.1979,
+      "step": 45820
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65474431428702e-05,
+      "loss": 1.2184,
+      "step": 45830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65466814934422e-05,
+      "loss": 1.1314,
+      "step": 45840
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65459198440142e-05,
+      "loss": 1.2293,
+      "step": 45850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65451581945862e-05,
+      "loss": 1.1224,
+      "step": 45860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6544396545158194e-05,
+      "loss": 1.0947,
+      "step": 45870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.65436348957302e-05,
+      "loss": 1.1553,
+      "step": 45880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6542873246302196e-05,
+      "loss": 1.0452,
+      "step": 45890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6542111596874194e-05,
+      "loss": 1.1067,
+      "step": 45900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.654134994744619e-05,
+      "loss": 1.1012,
+      "step": 45910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.654058829801819e-05,
+      "loss": 1.1354,
+      "step": 45920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653982664859019e-05,
+      "loss": 1.2844,
+      "step": 45930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653906499916218e-05,
+      "loss": 1.1399,
+      "step": 45940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653830334973419e-05,
+      "loss": 1.1722,
+      "step": 45950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6537541700306185e-05,
+      "loss": 1.0204,
+      "step": 45960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653678005087818e-05,
+      "loss": 1.1674,
+      "step": 45970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653601840145018e-05,
+      "loss": 1.0443,
+      "step": 45980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6535256752022184e-05,
+      "loss": 1.2304,
+      "step": 45990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6534495102594175e-05,
+      "loss": 1.1463,
+      "step": 46000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653373345316618e-05,
+      "loss": 1.1234,
+      "step": 46010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653297180373818e-05,
+      "loss": 1.1436,
+      "step": 46020
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6532210154310174e-05,
+      "loss": 1.1641,
+      "step": 46030
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.653144850488217e-05,
+      "loss": 1.06,
+      "step": 46040
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6530686855454176e-05,
+      "loss": 0.9795,
+      "step": 46050
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.652992520602617e-05,
+      "loss": 1.2983,
+      "step": 46060
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.652916355659817e-05,
+      "loss": 1.1872,
+      "step": 46070
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.652840190717017e-05,
+      "loss": 1.0974,
+      "step": 46080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6527640257742166e-05,
+      "loss": 1.159,
+      "step": 46090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.652687860831417e-05,
+      "loss": 1.1574,
+      "step": 46100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.652611695888617e-05,
+      "loss": 1.1521,
+      "step": 46110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6525355309458165e-05,
+      "loss": 1.0385,
+      "step": 46120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.652459366003016e-05,
+      "loss": 1.1206,
+      "step": 46130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6523832010602167e-05,
+      "loss": 1.1658,
+      "step": 46140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.652307036117416e-05,
+      "loss": 1.0668,
+      "step": 46150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.652230871174616e-05,
+      "loss": 1.0881,
+      "step": 46160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.652154706231816e-05,
+      "loss": 1.0558,
+      "step": 46170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6520785412890156e-05,
+      "loss": 1.1704,
+      "step": 46180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6520023763462154e-05,
+      "loss": 1.1326,
+      "step": 46190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651926211403416e-05,
+      "loss": 1.1523,
+      "step": 46200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651850046460615e-05,
+      "loss": 1.0912,
+      "step": 46210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651773881517815e-05,
+      "loss": 1.1827,
+      "step": 46220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651697716575015e-05,
+      "loss": 1.1651,
+      "step": 46230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651621551632215e-05,
+      "loss": 1.1059,
+      "step": 46240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6515453866894146e-05,
+      "loss": 1.0754,
+      "step": 46250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651469221746615e-05,
+      "loss": 1.0902,
+      "step": 46260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651393056803815e-05,
+      "loss": 1.1605,
+      "step": 46270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6513168918610145e-05,
+      "loss": 1.0847,
+      "step": 46280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651240726918214e-05,
+      "loss": 1.0435,
+      "step": 46290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.651164561975414e-05,
+      "loss": 1.2554,
+      "step": 46300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6510883970326144e-05,
+      "loss": 1.1524,
+      "step": 46310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.651012232089814e-05,
+      "loss": 1.1272,
+      "step": 46320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650936067147014e-05,
+      "loss": 1.1877,
+      "step": 46330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6508599022042136e-05,
+      "loss": 1.0833,
+      "step": 46340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650783737261414e-05,
+      "loss": 1.2337,
+      "step": 46350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650707572318613e-05,
+      "loss": 1.1231,
+      "step": 46360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6506314073758135e-05,
+      "loss": 0.9644,
+      "step": 46370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6505552424330126e-05,
+      "loss": 1.136,
+      "step": 46380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650479077490213e-05,
+      "loss": 1.189,
+      "step": 46390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650402912547413e-05,
+      "loss": 0.9831,
+      "step": 46400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6503267476046125e-05,
+      "loss": 1.0627,
+      "step": 46410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650250582661812e-05,
+      "loss": 1.1042,
+      "step": 46420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650174417719013e-05,
+      "loss": 1.1739,
+      "step": 46430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6500982527762124e-05,
+      "loss": 1.155,
+      "step": 46440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.650022087833412e-05,
+      "loss": 1.227,
+      "step": 46450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.649945922890612e-05,
+      "loss": 0.9736,
+      "step": 46460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.649869757947812e-05,
+      "loss": 1.0631,
+      "step": 46470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.649793593005012e-05,
+      "loss": 1.2706,
+      "step": 46480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.649717428062212e-05,
+      "loss": 1.0121,
+      "step": 46490
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6496412631194116e-05,
+      "loss": 1.0125,
+      "step": 46500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6495650981766114e-05,
+      "loss": 1.0327,
+      "step": 46510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.649488933233812e-05,
+      "loss": 1.0114,
+      "step": 46520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.649412768291011e-05,
+      "loss": 1.0914,
+      "step": 46530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.649336603348211e-05,
+      "loss": 1.0573,
+      "step": 46540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.649260438405411e-05,
+      "loss": 1.1207,
+      "step": 46550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.649184273462611e-05,
+      "loss": 0.9854,
+      "step": 46560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6491081085198105e-05,
+      "loss": 1.0961,
+      "step": 46570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.649031943577011e-05,
+      "loss": 1.0079,
+      "step": 46580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64895577863421e-05,
+      "loss": 1.1117,
+      "step": 46590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6488796136914104e-05,
+      "loss": 1.1937,
+      "step": 46600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64880344874861e-05,
+      "loss": 1.0454,
+      "step": 46610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64872728380581e-05,
+      "loss": 1.0029,
+      "step": 46620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64865111886301e-05,
+      "loss": 1.1099,
+      "step": 46630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64857495392021e-05,
+      "loss": 1.0713,
+      "step": 46640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64849878897741e-05,
+      "loss": 1.1292,
+      "step": 46650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6484226240346096e-05,
+      "loss": 1.1014,
+      "step": 46660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.648346459091809e-05,
+      "loss": 1.0986,
+      "step": 46670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.648270294149009e-05,
+      "loss": 0.9874,
+      "step": 46680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6481941292062095e-05,
+      "loss": 1.1313,
+      "step": 46690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.648117964263409e-05,
+      "loss": 1.0244,
+      "step": 46700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.648041799320609e-05,
+      "loss": 1.061,
+      "step": 46710
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.647965634377809e-05,
+      "loss": 0.9697,
+      "step": 46720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.647889469435009e-05,
+      "loss": 1.1313,
+      "step": 46730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.647813304492208e-05,
+      "loss": 1.03,
+      "step": 46740
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6477371395494087e-05,
+      "loss": 1.086,
+      "step": 46750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6476609746066084e-05,
+      "loss": 1.1556,
+      "step": 46760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.647584809663808e-05,
+      "loss": 1.0368,
+      "step": 46770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.647508644721008e-05,
+      "loss": 1.0437,
+      "step": 46780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.647432479778208e-05,
+      "loss": 1.1291,
+      "step": 46790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6473563148354074e-05,
+      "loss": 0.9685,
+      "step": 46800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.647280149892608e-05,
+      "loss": 1.2006,
+      "step": 46810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6472039849498076e-05,
+      "loss": 1.1038,
+      "step": 46820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.647127820007007e-05,
+      "loss": 1.1435,
+      "step": 46830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.647051655064207e-05,
+      "loss": 1.0309,
+      "step": 46840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646975490121407e-05,
+      "loss": 0.9889,
+      "step": 46850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646899325178607e-05,
+      "loss": 1.1656,
+      "step": 46860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646823160235807e-05,
+      "loss": 0.898,
+      "step": 46870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646746995293007e-05,
+      "loss": 0.9914,
+      "step": 46880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6466708303502065e-05,
+      "loss": 1.183,
+      "step": 46890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646594665407407e-05,
+      "loss": 1.0018,
+      "step": 46900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646518500464606e-05,
+      "loss": 1.159,
+      "step": 46910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6464423355218064e-05,
+      "loss": 1.1358,
+      "step": 46920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646366170579006e-05,
+      "loss": 1.1217,
+      "step": 46930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646290005636206e-05,
+      "loss": 1.1379,
+      "step": 46940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6462138406934056e-05,
+      "loss": 1.1009,
+      "step": 46950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646137675750606e-05,
+      "loss": 1.1613,
+      "step": 46960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.646061510807805e-05,
+      "loss": 1.1047,
+      "step": 46970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6459853458650055e-05,
+      "loss": 1.0796,
+      "step": 46980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645909180922205e-05,
+      "loss": 1.1647,
+      "step": 46990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645833015979405e-05,
+      "loss": 1.0942,
+      "step": 47000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645756851036605e-05,
+      "loss": 1.0369,
+      "step": 47010
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645680686093805e-05,
+      "loss": 1.0728,
+      "step": 47020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645604521151005e-05,
+      "loss": 1.0236,
+      "step": 47030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645528356208205e-05,
+      "loss": 1.1434,
+      "step": 47040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6454521912654044e-05,
+      "loss": 1.1037,
+      "step": 47050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645376026322604e-05,
+      "loss": 1.0539,
+      "step": 47060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6452998613798046e-05,
+      "loss": 1.0723,
+      "step": 47070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6452236964370044e-05,
+      "loss": 1.0698,
+      "step": 47080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645147531494204e-05,
+      "loss": 1.1568,
+      "step": 47090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.645071366551404e-05,
+      "loss": 1.1255,
+      "step": 47100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.644995201608604e-05,
+      "loss": 1.1177,
+      "step": 47110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6449190366658034e-05,
+      "loss": 1.1521,
+      "step": 47120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.644842871723004e-05,
+      "loss": 1.1237,
+      "step": 47130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6447667067802035e-05,
+      "loss": 1.0977,
+      "step": 47140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.644690541837403e-05,
+      "loss": 1.1022,
+      "step": 47150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.644614376894603e-05,
+      "loss": 1.0735,
+      "step": 47160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6445382119518034e-05,
+      "loss": 1.0003,
+      "step": 47170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6444620470090025e-05,
+      "loss": 1.0842,
+      "step": 47180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.644385882066203e-05,
+      "loss": 1.015,
+      "step": 47190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.644309717123403e-05,
+      "loss": 0.9956,
+      "step": 47200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6442335521806024e-05,
+      "loss": 0.9721,
+      "step": 47210
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.644157387237802e-05,
+      "loss": 1.0662,
+      "step": 47220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6440812222950026e-05,
+      "loss": 1.0836,
+      "step": 47230
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6440050573522023e-05,
+      "loss": 1.2038,
+      "step": 47240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643928892409402e-05,
+      "loss": 1.1651,
+      "step": 47250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643852727466602e-05,
+      "loss": 0.9752,
+      "step": 47260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6437765625238016e-05,
+      "loss": 1.1217,
+      "step": 47270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643700397581002e-05,
+      "loss": 1.0966,
+      "step": 47280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643624232638202e-05,
+      "loss": 1.2025,
+      "step": 47290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6435480676954015e-05,
+      "loss": 1.0723,
+      "step": 47300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643471902752601e-05,
+      "loss": 1.1373,
+      "step": 47310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643395737809801e-05,
+      "loss": 1.1228,
+      "step": 47320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643319572867001e-05,
+      "loss": 1.2033,
+      "step": 47330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643243407924201e-05,
+      "loss": 1.2877,
+      "step": 47340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6431672429814e-05,
+      "loss": 1.0933,
+      "step": 47350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643091078038601e-05,
+      "loss": 1.1986,
+      "step": 47360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6430149130958004e-05,
+      "loss": 1.1484,
+      "step": 47370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.642938748153e-05,
+      "loss": 1.0959,
+      "step": 47380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6428625832102e-05,
+      "loss": 1.0715,
+      "step": 47390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6427864182674e-05,
+      "loss": 1.1617,
+      "step": 47400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6427102533246e-05,
+      "loss": 1.2855,
+      "step": 47410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6426340883818e-05,
+      "loss": 0.9895,
+      "step": 47420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6425579234389996e-05,
+      "loss": 1.0981,
+      "step": 47430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.642481758496199e-05,
+      "loss": 1.1212,
+      "step": 47440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6424055935534e-05,
+      "loss": 1.1814,
+      "step": 47450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6423294286105995e-05,
+      "loss": 1.085,
+      "step": 47460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.642253263667799e-05,
+      "loss": 1.0087,
+      "step": 47470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.642177098724999e-05,
+      "loss": 1.1489,
+      "step": 47480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6421009337821994e-05,
+      "loss": 1.2351,
+      "step": 47490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6420247688393985e-05,
+      "loss": 1.2329,
+      "step": 47500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641948603896599e-05,
+      "loss": 1.1096,
+      "step": 47510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6418724389537986e-05,
+      "loss": 1.086,
+      "step": 47520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6417962740109984e-05,
+      "loss": 1.2014,
+      "step": 47530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641720109068198e-05,
+      "loss": 1.0575,
+      "step": 47540
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6416439441253986e-05,
+      "loss": 1.0042,
+      "step": 47550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6415677791825976e-05,
+      "loss": 1.1868,
+      "step": 47560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641491614239798e-05,
+      "loss": 1.0174,
+      "step": 47570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641415449296998e-05,
+      "loss": 1.0917,
+      "step": 47580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6413392843541975e-05,
+      "loss": 1.1115,
+      "step": 47590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641263119411397e-05,
+      "loss": 1.0918,
+      "step": 47600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641186954468598e-05,
+      "loss": 1.1192,
+      "step": 47610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6411107895257975e-05,
+      "loss": 1.1189,
+      "step": 47620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.641034624582997e-05,
+      "loss": 1.1584,
+      "step": 47630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640958459640197e-05,
+      "loss": 1.0525,
+      "step": 47640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640882294697397e-05,
+      "loss": 1.1129,
+      "step": 47650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640806129754597e-05,
+      "loss": 1.115,
+      "step": 47660
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640729964811797e-05,
+      "loss": 1.0803,
+      "step": 47670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6406537998689966e-05,
+      "loss": 1.0646,
+      "step": 47680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6405776349261964e-05,
+      "loss": 1.2251,
+      "step": 47690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640501469983397e-05,
+      "loss": 1.0686,
+      "step": 47700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640425305040596e-05,
+      "loss": 1.1943,
+      "step": 47710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640349140097796e-05,
+      "loss": 1.1466,
+      "step": 47720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640272975154996e-05,
+      "loss": 1.119,
+      "step": 47730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640196810212196e-05,
+      "loss": 1.0764,
+      "step": 47740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6401206452693955e-05,
+      "loss": 0.9639,
+      "step": 47750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.640044480326595e-05,
+      "loss": 1.1252,
+      "step": 47760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.639968315383795e-05,
+      "loss": 1.0808,
+      "step": 47770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6398921504409954e-05,
+      "loss": 1.1903,
+      "step": 47780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6398159854981945e-05,
+      "loss": 1.1875,
+      "step": 47790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.639739820555395e-05,
+      "loss": 1.1115,
+      "step": 47800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.639663655612595e-05,
+      "loss": 1.0716,
+      "step": 47810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6395874906697944e-05,
+      "loss": 1.1113,
+      "step": 47820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.639511325726995e-05,
+      "loss": 1.0571,
+      "step": 47830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6394351607841946e-05,
+      "loss": 1.1889,
+      "step": 47840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6393589958413943e-05,
+      "loss": 1.0816,
+      "step": 47850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.639282830898594e-05,
+      "loss": 1.175,
+      "step": 47860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6392066659557945e-05,
+      "loss": 1.1129,
+      "step": 47870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6391305010129936e-05,
+      "loss": 1.1004,
+      "step": 47880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.639054336070194e-05,
+      "loss": 1.0535,
+      "step": 47890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638978171127394e-05,
+      "loss": 1.1536,
+      "step": 47900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6389020061845935e-05,
+      "loss": 1.1066,
+      "step": 47910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638825841241793e-05,
+      "loss": 1.0203,
+      "step": 47920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638749676298994e-05,
+      "loss": 1.2252,
+      "step": 47930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638673511356193e-05,
+      "loss": 1.0769,
+      "step": 47940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638597346413393e-05,
+      "loss": 1.0666,
+      "step": 47950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638521181470593e-05,
+      "loss": 1.0886,
+      "step": 47960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638445016527793e-05,
+      "loss": 1.1259,
+      "step": 47970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6383688515849924e-05,
+      "loss": 1.0874,
+      "step": 47980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638292686642193e-05,
+      "loss": 1.1349,
+      "step": 47990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6382165216993926e-05,
+      "loss": 1.1867,
+      "step": 48000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638140356756592e-05,
+      "loss": 1.1291,
+      "step": 48010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.638064191813792e-05,
+      "loss": 1.2372,
+      "step": 48020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.637988026870992e-05,
+      "loss": 1.2784,
+      "step": 48030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.637911861928192e-05,
+      "loss": 1.161,
+      "step": 48040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.637835696985392e-05,
+      "loss": 1.0279,
+      "step": 48050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.637759532042592e-05,
+      "loss": 1.1091,
+      "step": 48060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6376833670997915e-05,
+      "loss": 1.277,
+      "step": 48070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.637607202156992e-05,
+      "loss": 1.0235,
+      "step": 48080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.637531037214191e-05,
+      "loss": 1.0531,
+      "step": 48090
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6374548722713914e-05,
+      "loss": 1.1754,
+      "step": 48100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.637378707328591e-05,
+      "loss": 1.0585,
+      "step": 48110
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.637302542385791e-05,
+      "loss": 1.1441,
+      "step": 48120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6372263774429906e-05,
+      "loss": 1.1438,
+      "step": 48130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.637150212500191e-05,
+      "loss": 1.1529,
+      "step": 48140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63707404755739e-05,
+      "loss": 1.1465,
+      "step": 48150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6369978826145906e-05,
+      "loss": 1.0558,
+      "step": 48160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63692171767179e-05,
+      "loss": 1.0581,
+      "step": 48170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63684555272899e-05,
+      "loss": 1.0262,
+      "step": 48180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63676938778619e-05,
+      "loss": 1.136,
+      "step": 48190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63669322284339e-05,
+      "loss": 1.1,
+      "step": 48200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63661705790059e-05,
+      "loss": 1.1876,
+      "step": 48210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63654089295779e-05,
+      "loss": 1.1806,
+      "step": 48220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6364647280149895e-05,
+      "loss": 1.0973,
+      "step": 48230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.636388563072189e-05,
+      "loss": 1.2446,
+      "step": 48240
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6363123981293896e-05,
+      "loss": 1.1792,
+      "step": 48250
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.636236233186589e-05,
+      "loss": 1.1988,
+      "step": 48260
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.636160068243789e-05,
+      "loss": 1.0704,
+      "step": 48270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.636083903300989e-05,
+      "loss": 1.0274,
+      "step": 48280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6360077383581886e-05,
+      "loss": 1.052,
+      "step": 48290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6359315734153884e-05,
+      "loss": 1.3271,
+      "step": 48300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635855408472589e-05,
+      "loss": 1.1324,
+      "step": 48310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635779243529788e-05,
+      "loss": 1.0539,
+      "step": 48320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635703078586988e-05,
+      "loss": 1.1184,
+      "step": 48330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635626913644188e-05,
+      "loss": 1.0074,
+      "step": 48340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635550748701388e-05,
+      "loss": 1.1069,
+      "step": 48350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6354745837585875e-05,
+      "loss": 1.1274,
+      "step": 48360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635398418815788e-05,
+      "loss": 1.1089,
+      "step": 48370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635322253872987e-05,
+      "loss": 1.204,
+      "step": 48380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6352460889301874e-05,
+      "loss": 1.0702,
+      "step": 48390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635169923987387e-05,
+      "loss": 1.0603,
+      "step": 48400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.635093759044587e-05,
+      "loss": 1.0433,
+      "step": 48410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6350175941017874e-05,
+      "loss": 1.0749,
+      "step": 48420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.634941429158987e-05,
+      "loss": 1.1426,
+      "step": 48430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.634865264216187e-05,
+      "loss": 1.0933,
+      "step": 48440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6347890992733866e-05,
+      "loss": 1.1238,
+      "step": 48450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.634712934330587e-05,
+      "loss": 1.0234,
+      "step": 48460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.634636769387786e-05,
+      "loss": 1.0565,
+      "step": 48470
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6345606044449865e-05,
+      "loss": 1.1986,
+      "step": 48480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.634484439502186e-05,
+      "loss": 1.0894,
+      "step": 48490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.634408274559386e-05,
+      "loss": 0.9925,
+      "step": 48500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.634332109616586e-05,
+      "loss": 0.9884,
+      "step": 48510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.634255944673786e-05,
+      "loss": 1.1059,
+      "step": 48520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.634179779730985e-05,
+      "loss": 1.0278,
+      "step": 48530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.634103614788186e-05,
+      "loss": 1.0513,
+      "step": 48540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6340274498453854e-05,
+      "loss": 1.0477,
+      "step": 48550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633951284902585e-05,
+      "loss": 1.2569,
+      "step": 48560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633875119959785e-05,
+      "loss": 1.1628,
+      "step": 48570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6337989550169853e-05,
+      "loss": 1.1645,
+      "step": 48580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633722790074185e-05,
+      "loss": 1.0527,
+      "step": 48590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633646625131385e-05,
+      "loss": 0.9288,
+      "step": 48600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6335704601885846e-05,
+      "loss": 1.014,
+      "step": 48610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633494295245784e-05,
+      "loss": 1.176,
+      "step": 48620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633418130302985e-05,
+      "loss": 1.1414,
+      "step": 48630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6333419653601845e-05,
+      "loss": 1.1055,
+      "step": 48640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633265800417384e-05,
+      "loss": 1.0651,
+      "step": 48650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.633189635474584e-05,
+      "loss": 1.0736,
+      "step": 48660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6331134705317844e-05,
+      "loss": 1.0633,
+      "step": 48670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6330373055889835e-05,
+      "loss": 1.2007,
+      "step": 48680
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.632961140646184e-05,
+      "loss": 1.208,
+      "step": 48690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.632884975703383e-05,
+      "loss": 1.0834,
+      "step": 48700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6328088107605834e-05,
+      "loss": 1.0695,
+      "step": 48710
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.632732645817783e-05,
+      "loss": 0.9941,
+      "step": 48720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.632656480874983e-05,
+      "loss": 0.9525,
+      "step": 48730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6325803159321826e-05,
+      "loss": 1.0966,
+      "step": 48740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.632504150989383e-05,
+      "loss": 1.1353,
+      "step": 48750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.632427986046582e-05,
+      "loss": 1.0943,
+      "step": 48760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6323518211037826e-05,
+      "loss": 1.0048,
+      "step": 48770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.632275656160982e-05,
+      "loss": 1.0883,
+      "step": 48780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.632199491218182e-05,
+      "loss": 1.1945,
+      "step": 48790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6321233262753825e-05,
+      "loss": 1.2022,
+      "step": 48800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.632047161332582e-05,
+      "loss": 1.1882,
+      "step": 48810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631970996389782e-05,
+      "loss": 1.1664,
+      "step": 48820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631894831446982e-05,
+      "loss": 1.0235,
+      "step": 48830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631818666504182e-05,
+      "loss": 1.0845,
+      "step": 48840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631742501561381e-05,
+      "loss": 1.1245,
+      "step": 48850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6316663366185816e-05,
+      "loss": 1.1892,
+      "step": 48860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6315901716757814e-05,
+      "loss": 1.1187,
+      "step": 48870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631514006732981e-05,
+      "loss": 1.0034,
+      "step": 48880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631437841790181e-05,
+      "loss": 1.1152,
+      "step": 48890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631361676847381e-05,
+      "loss": 0.9968,
+      "step": 48900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6312855119045804e-05,
+      "loss": 1.1177,
+      "step": 48910
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.631209346961781e-05,
+      "loss": 1.0943,
+      "step": 48920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6311331820189805e-05,
+      "loss": 1.0852,
+      "step": 48930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.63105701707618e-05,
+      "loss": 1.0316,
+      "step": 48940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.63098085213338e-05,
+      "loss": 1.1989,
+      "step": 48950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6309046871905805e-05,
+      "loss": 1.1181,
+      "step": 48960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6308285222477795e-05,
+      "loss": 1.0403,
+      "step": 48970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.63075235730498e-05,
+      "loss": 1.1422,
+      "step": 48980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.63067619236218e-05,
+      "loss": 0.9989,
+      "step": 48990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6306000274193794e-05,
+      "loss": 1.0852,
+      "step": 49000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.63052386247658e-05,
+      "loss": 1.0694,
+      "step": 49010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6304476975337796e-05,
+      "loss": 1.2272,
+      "step": 49020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6303715325909794e-05,
+      "loss": 1.0554,
+      "step": 49030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.630295367648179e-05,
+      "loss": 1.121,
+      "step": 49040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6302192027053795e-05,
+      "loss": 1.0362,
+      "step": 49050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6301430377625786e-05,
+      "loss": 1.2298,
+      "step": 49060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.630066872819779e-05,
+      "loss": 1.3405,
+      "step": 49070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629990707876979e-05,
+      "loss": 1.2117,
+      "step": 49080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6299145429341785e-05,
+      "loss": 1.0543,
+      "step": 49090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629838377991378e-05,
+      "loss": 1.1176,
+      "step": 49100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629762213048579e-05,
+      "loss": 1.0389,
+      "step": 49110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629686048105778e-05,
+      "loss": 1.1602,
+      "step": 49120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629609883162978e-05,
+      "loss": 1.0385,
+      "step": 49130
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629533718220177e-05,
+      "loss": 1.0945,
+      "step": 49140
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.629457553277378e-05,
+      "loss": 1.0565,
+      "step": 49150
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6293813883345774e-05,
+      "loss": 1.0134,
+      "step": 49160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.629305223391777e-05,
+      "loss": 1.1068,
+      "step": 49170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.629229058448977e-05,
+      "loss": 1.1165,
+      "step": 49180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6291528935061773e-05,
+      "loss": 1.1171,
+      "step": 49190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.629076728563377e-05,
+      "loss": 1.1589,
+      "step": 49200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.629000563620577e-05,
+      "loss": 1.115,
+      "step": 49210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628924398677777e-05,
+      "loss": 1.1793,
+      "step": 49220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628848233734976e-05,
+      "loss": 1.1605,
+      "step": 49230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628772068792177e-05,
+      "loss": 1.0926,
+      "step": 49240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6286959038493765e-05,
+      "loss": 1.1026,
+      "step": 49250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628619738906576e-05,
+      "loss": 1.0298,
+      "step": 49260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628543573963776e-05,
+      "loss": 1.1025,
+      "step": 49270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6284674090209764e-05,
+      "loss": 1.0944,
+      "step": 49280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6283912440781755e-05,
+      "loss": 1.1837,
+      "step": 49290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628315079135376e-05,
+      "loss": 0.9784,
+      "step": 49300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6282389141925757e-05,
+      "loss": 1.0972,
+      "step": 49310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6281627492497754e-05,
+      "loss": 0.998,
+      "step": 49320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.628086584306975e-05,
+      "loss": 1.0472,
+      "step": 49330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6280104193641756e-05,
+      "loss": 1.0538,
+      "step": 49340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6279342544213746e-05,
+      "loss": 1.0991,
+      "step": 49350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.627858089478575e-05,
+      "loss": 1.0813,
+      "step": 49360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.627781924535775e-05,
+      "loss": 0.9693,
+      "step": 49370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6277057595929746e-05,
+      "loss": 1.074,
+      "step": 49380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.627629594650175e-05,
+      "loss": 1.0672,
+      "step": 49390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.627553429707375e-05,
+      "loss": 0.9873,
+      "step": 49400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6274772647645745e-05,
+      "loss": 1.12,
+      "step": 49410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.627401099821774e-05,
+      "loss": 1.0613,
+      "step": 49420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6273249348789747e-05,
+      "loss": 1.0541,
+      "step": 49430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.627248769936174e-05,
+      "loss": 1.063,
+      "step": 49440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.627172604993374e-05,
+      "loss": 0.9295,
+      "step": 49450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.627096440050574e-05,
+      "loss": 0.9982,
+      "step": 49460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6270202751077736e-05,
+      "loss": 1.1775,
+      "step": 49470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6269441101649734e-05,
+      "loss": 1.0717,
+      "step": 49480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626867945222174e-05,
+      "loss": 1.1113,
+      "step": 49490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626791780279373e-05,
+      "loss": 1.148,
+      "step": 49500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626715615336573e-05,
+      "loss": 1.0384,
+      "step": 49510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626639450393773e-05,
+      "loss": 1.0338,
+      "step": 49520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626563285450973e-05,
+      "loss": 1.1486,
+      "step": 49530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6264871205081725e-05,
+      "loss": 1.1878,
+      "step": 49540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626410955565373e-05,
+      "loss": 1.0586,
+      "step": 49550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626334790622572e-05,
+      "loss": 0.932,
+      "step": 49560
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6262586256797725e-05,
+      "loss": 1.1312,
+      "step": 49570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626182460736972e-05,
+      "loss": 1.1794,
+      "step": 49580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.626106295794172e-05,
+      "loss": 1.1549,
+      "step": 49590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6260301308513724e-05,
+      "loss": 1.0794,
+      "step": 49600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6259539659085714e-05,
+      "loss": 1.1949,
+      "step": 49610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.625877800965772e-05,
+      "loss": 1.0724,
+      "step": 49620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6258016360229716e-05,
+      "loss": 1.0494,
+      "step": 49630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6257254710801714e-05,
+      "loss": 1.1551,
+      "step": 49640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.625649306137371e-05,
+      "loss": 1.2162,
+      "step": 49650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6255731411945715e-05,
+      "loss": 1.0598,
+      "step": 49660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6254969762517706e-05,
+      "loss": 1.2138,
+      "step": 49670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.625420811308971e-05,
+      "loss": 1.0913,
+      "step": 49680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.625344646366171e-05,
+      "loss": 1.162,
+      "step": 49690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6252684814233705e-05,
+      "loss": 1.0096,
+      "step": 49700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62519231648057e-05,
+      "loss": 1.0614,
+      "step": 49710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.625116151537771e-05,
+      "loss": 1.0875,
+      "step": 49720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62503998659497e-05,
+      "loss": 1.0723,
+      "step": 49730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62496382165217e-05,
+      "loss": 0.9348,
+      "step": 49740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62488765670937e-05,
+      "loss": 1.205,
+      "step": 49750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62481149176657e-05,
+      "loss": 1.0686,
+      "step": 49760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6247353268237694e-05,
+      "loss": 1.0153,
+      "step": 49770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62465916188097e-05,
+      "loss": 1.2193,
+      "step": 49780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6245829969381696e-05,
+      "loss": 1.1266,
+      "step": 49790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6245068319953693e-05,
+      "loss": 1.055,
+      "step": 49800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62443066705257e-05,
+      "loss": 1.0358,
+      "step": 49810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.624354502109769e-05,
+      "loss": 1.0456,
+      "step": 49820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.624278337166969e-05,
+      "loss": 1.0626,
+      "step": 49830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.624202172224169e-05,
+      "loss": 0.9967,
+      "step": 49840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.624126007281369e-05,
+      "loss": 1.0593,
+      "step": 49850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6240498423385685e-05,
+      "loss": 1.2882,
+      "step": 49860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623973677395769e-05,
+      "loss": 1.1106,
+      "step": 49870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623897512452968e-05,
+      "loss": 1.0327,
+      "step": 49880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6238213475101684e-05,
+      "loss": 0.9992,
+      "step": 49890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623745182567368e-05,
+      "loss": 1.169,
+      "step": 49900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623669017624568e-05,
+      "loss": 1.1006,
+      "step": 49910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6235928526817677e-05,
+      "loss": 1.0323,
+      "step": 49920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623516687738968e-05,
+      "loss": 1.0988,
+      "step": 49930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623440522796167e-05,
+      "loss": 1.1676,
+      "step": 49940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6233643578533676e-05,
+      "loss": 1.1215,
+      "step": 49950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623288192910567e-05,
+      "loss": 1.1104,
+      "step": 49960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623212027967767e-05,
+      "loss": 1.1723,
+      "step": 49970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623135863024967e-05,
+      "loss": 1.0501,
+      "step": 49980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.623059698082167e-05,
+      "loss": 1.0473,
+      "step": 49990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.622983533139367e-05,
+      "loss": 1.2386,
+      "step": 50000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.622907368196567e-05,
+      "loss": 1.1626,
+      "step": 50010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.622831203253767e-05,
+      "loss": 1.028,
+      "step": 50020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.622755038310966e-05,
+      "loss": 1.0974,
+      "step": 50030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6226788733681667e-05,
+      "loss": 1.0405,
+      "step": 50040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6226027084253664e-05,
+      "loss": 1.0198,
+      "step": 50050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.622526543482566e-05,
+      "loss": 1.0152,
+      "step": 50060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.622450378539766e-05,
+      "loss": 1.1674,
+      "step": 50070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6223742135969656e-05,
+      "loss": 1.2915,
+      "step": 50080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6222980486541654e-05,
+      "loss": 1.1995,
+      "step": 50090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.622221883711366e-05,
+      "loss": 0.9781,
+      "step": 50100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.622145718768565e-05,
+      "loss": 1.0306,
+      "step": 50110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.622069553825765e-05,
+      "loss": 1.0411,
+      "step": 50120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621993388882965e-05,
+      "loss": 1.0332,
+      "step": 50130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621917223940165e-05,
+      "loss": 1.1308,
+      "step": 50140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6218410589973645e-05,
+      "loss": 1.1031,
+      "step": 50150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621764894054565e-05,
+      "loss": 1.0729,
+      "step": 50160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621688729111765e-05,
+      "loss": 1.1961,
+      "step": 50170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6216125641689645e-05,
+      "loss": 1.1264,
+      "step": 50180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621536399226165e-05,
+      "loss": 1.1762,
+      "step": 50190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621460234283364e-05,
+      "loss": 1.0514,
+      "step": 50200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6213840693405644e-05,
+      "loss": 1.2695,
+      "step": 50210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621307904397764e-05,
+      "loss": 1.0713,
+      "step": 50220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621231739454964e-05,
+      "loss": 0.9875,
+      "step": 50230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6211555745121636e-05,
+      "loss": 1.1564,
+      "step": 50240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.621079409569364e-05,
+      "loss": 1.0913,
+      "step": 50250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.621003244626563e-05,
+      "loss": 1.2448,
+      "step": 50260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6209270796837635e-05,
+      "loss": 1.1585,
+      "step": 50270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620850914740963e-05,
+      "loss": 0.9876,
+      "step": 50280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620774749798163e-05,
+      "loss": 1.0442,
+      "step": 50290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620698584855363e-05,
+      "loss": 1.1275,
+      "step": 50300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620622419912563e-05,
+      "loss": 1.0757,
+      "step": 50310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620546254969762e-05,
+      "loss": 1.1791,
+      "step": 50320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620470090026963e-05,
+      "loss": 0.9599,
+      "step": 50330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6203939250841624e-05,
+      "loss": 1.0275,
+      "step": 50340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620317760141362e-05,
+      "loss": 1.0337,
+      "step": 50350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620241595198562e-05,
+      "loss": 1.2931,
+      "step": 50360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6201654302557624e-05,
+      "loss": 1.1051,
+      "step": 50370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620089265312962e-05,
+      "loss": 1.1608,
+      "step": 50380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.620013100370162e-05,
+      "loss": 1.1564,
+      "step": 50390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.619936935427362e-05,
+      "loss": 0.9763,
+      "step": 50400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6198607704845613e-05,
+      "loss": 1.0511,
+      "step": 50410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.619784605541762e-05,
+      "loss": 1.1002,
+      "step": 50420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6197084405989615e-05,
+      "loss": 1.0519,
+      "step": 50430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.619632275656161e-05,
+      "loss": 1.1224,
+      "step": 50440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.619556110713361e-05,
+      "loss": 1.1359,
+      "step": 50450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6194799457705614e-05,
+      "loss": 1.0844,
+      "step": 50460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6194037808277605e-05,
+      "loss": 1.1983,
+      "step": 50470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.619327615884961e-05,
+      "loss": 1.1128,
+      "step": 50480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.619251450942161e-05,
+      "loss": 1.0661,
+      "step": 50490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6191752859993604e-05,
+      "loss": 1.1324,
+      "step": 50500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.61909912105656e-05,
+      "loss": 1.1022,
+      "step": 50510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.61902295611376e-05,
+      "loss": 1.0605,
+      "step": 50520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6189467911709597e-05,
+      "loss": 1.1815,
+      "step": 50530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.61887062622816e-05,
+      "loss": 1.1301,
+      "step": 50540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.61879446128536e-05,
+      "loss": 1.1313,
+      "step": 50550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6187182963425596e-05,
+      "loss": 1.1014,
+      "step": 50560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618642131399759e-05,
+      "loss": 1.0808,
+      "step": 50570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618565966456959e-05,
+      "loss": 1.1129,
+      "step": 50580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6184898015141595e-05,
+      "loss": 1.0201,
+      "step": 50590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618413636571359e-05,
+      "loss": 1.1531,
+      "step": 50600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618337471628559e-05,
+      "loss": 1.1585,
+      "step": 50610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618261306685759e-05,
+      "loss": 1.0835,
+      "step": 50620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618185141742959e-05,
+      "loss": 1.0176,
+      "step": 50630
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618108976800158e-05,
+      "loss": 1.096,
+      "step": 50640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6180328118573587e-05,
+      "loss": 1.1365,
+      "step": 50650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6179566469145584e-05,
+      "loss": 1.0347,
+      "step": 50660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.617880481971758e-05,
+      "loss": 1.0818,
+      "step": 50670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.617804317028958e-05,
+      "loss": 1.2149,
+      "step": 50680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.617728152086158e-05,
+      "loss": 1.14,
+      "step": 50690
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6176519871433574e-05,
+      "loss": 1.1775,
+      "step": 50700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.617575822200558e-05,
+      "loss": 0.9993,
+      "step": 50710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6174996572577576e-05,
+      "loss": 1.0802,
+      "step": 50720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.617423492314957e-05,
+      "loss": 1.2363,
+      "step": 50730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.617347327372157e-05,
+      "loss": 1.1647,
+      "step": 50740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6172711624293575e-05,
+      "loss": 1.1489,
+      "step": 50750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.617194997486557e-05,
+      "loss": 1.1038,
+      "step": 50760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.617118832543757e-05,
+      "loss": 1.1618,
+      "step": 50770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.617042667600957e-05,
+      "loss": 1.0758,
+      "step": 50780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6169665026581565e-05,
+      "loss": 1.1022,
+      "step": 50790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616890337715357e-05,
+      "loss": 1.0762,
+      "step": 50800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6168141727725566e-05,
+      "loss": 1.193,
+      "step": 50810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6167380078297564e-05,
+      "loss": 1.1127,
+      "step": 50820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616661842886956e-05,
+      "loss": 1.1175,
+      "step": 50830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6165856779441566e-05,
+      "loss": 1.0406,
+      "step": 50840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6165095130013556e-05,
+      "loss": 1.0422,
+      "step": 50850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616433348058556e-05,
+      "loss": 1.1073,
+      "step": 50860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616357183115756e-05,
+      "loss": 1.1418,
+      "step": 50870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6162810181729555e-05,
+      "loss": 1.1621,
+      "step": 50880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616204853230155e-05,
+      "loss": 1.1683,
+      "step": 50890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616128688287356e-05,
+      "loss": 1.07,
+      "step": 50900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.616052523344555e-05,
+      "loss": 1.0215,
+      "step": 50910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615976358401755e-05,
+      "loss": 1.1894,
+      "step": 50920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615900193458955e-05,
+      "loss": 1.03,
+      "step": 50930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615824028516155e-05,
+      "loss": 1.1964,
+      "step": 50940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6157478635733544e-05,
+      "loss": 1.0965,
+      "step": 50950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615671698630555e-05,
+      "loss": 1.0164,
+      "step": 50960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6155955336877546e-05,
+      "loss": 1.0702,
+      "step": 50970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6155193687449544e-05,
+      "loss": 1.0321,
+      "step": 50980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615443203802154e-05,
+      "loss": 1.1099,
+      "step": 50990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615367038859354e-05,
+      "loss": 1.0287,
+      "step": 51000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615290873916554e-05,
+      "loss": 0.9338,
+      "step": 51010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6152147089737533e-05,
+      "loss": 1.1939,
+      "step": 51020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615138544030954e-05,
+      "loss": 1.1126,
+      "step": 51030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6150623790881535e-05,
+      "loss": 1.2771,
+      "step": 51040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.614986214145353e-05,
+      "loss": 1.1267,
+      "step": 51050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.614910049202553e-05,
+      "loss": 1.0585,
+      "step": 51060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6148338842597534e-05,
+      "loss": 1.1687,
+      "step": 51070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6147577193169525e-05,
+      "loss": 1.0697,
+      "step": 51080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.614681554374153e-05,
+      "loss": 1.133,
+      "step": 51090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.614605389431353e-05,
+      "loss": 0.9748,
+      "step": 51100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6145292244885524e-05,
+      "loss": 0.9818,
+      "step": 51110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.614453059545752e-05,
+      "loss": 1.0759,
+      "step": 51120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6143768946029526e-05,
+      "loss": 0.9642,
+      "step": 51130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6143007296601523e-05,
+      "loss": 1.1183,
+      "step": 51140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.614224564717352e-05,
+      "loss": 1.0601,
+      "step": 51150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.614148399774552e-05,
+      "loss": 1.2023,
+      "step": 51160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6140722348317516e-05,
+      "loss": 1.119,
+      "step": 51170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613996069888952e-05,
+      "loss": 0.9838,
+      "step": 51180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613919904946152e-05,
+      "loss": 1.0525,
+      "step": 51190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6138437400033515e-05,
+      "loss": 1.1203,
+      "step": 51200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613767575060551e-05,
+      "loss": 1.0854,
+      "step": 51210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613691410117752e-05,
+      "loss": 1.2052,
+      "step": 51220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613615245174951e-05,
+      "loss": 1.0817,
+      "step": 51230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613539080232151e-05,
+      "loss": 1.0913,
+      "step": 51240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613462915289351e-05,
+      "loss": 1.059,
+      "step": 51250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6133867503465507e-05,
+      "loss": 1.0723,
+      "step": 51260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6133105854037504e-05,
+      "loss": 1.06,
+      "step": 51270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.613234420460951e-05,
+      "loss": 1.0855,
+      "step": 51280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61315825551815e-05,
+      "loss": 1.1071,
+      "step": 51290
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61308209057535e-05,
+      "loss": 1.2184,
+      "step": 51300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61300592563255e-05,
+      "loss": 1.1301,
+      "step": 51310
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61292976068975e-05,
+      "loss": 1.0831,
+      "step": 51320
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6128535957469496e-05,
+      "loss": 1.1053,
+      "step": 51330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61277743080415e-05,
+      "loss": 1.1354,
+      "step": 51340
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61270126586135e-05,
+      "loss": 1.2167,
+      "step": 51350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6126251009185495e-05,
+      "loss": 1.1098,
+      "step": 51360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.612548935975749e-05,
+      "loss": 1.1077,
+      "step": 51370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.612472771032949e-05,
+      "loss": 1.0298,
+      "step": 51380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6123966060901494e-05,
+      "loss": 1.0476,
+      "step": 51390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.612320441147349e-05,
+      "loss": 1.0258,
+      "step": 51400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.612244276204549e-05,
+      "loss": 0.9867,
+      "step": 51410
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6121681112617486e-05,
+      "loss": 1.0194,
+      "step": 51420
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.612091946318949e-05,
+      "loss": 1.0539,
+      "step": 51430
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.612015781376148e-05,
+      "loss": 1.1079,
+      "step": 51440
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6119396164333486e-05,
+      "loss": 1.2019,
+      "step": 51450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6118634514905476e-05,
+      "loss": 1.0031,
+      "step": 51460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611787286547748e-05,
+      "loss": 1.2173,
+      "step": 51470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611711121604948e-05,
+      "loss": 1.0112,
+      "step": 51480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6116349566621475e-05,
+      "loss": 1.1729,
+      "step": 51490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611558791719347e-05,
+      "loss": 1.0465,
+      "step": 51500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611482626776548e-05,
+      "loss": 1.2137,
+      "step": 51510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6114064618337475e-05,
+      "loss": 1.0566,
+      "step": 51520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611330296890947e-05,
+      "loss": 1.047,
+      "step": 51530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611254131948147e-05,
+      "loss": 1.2399,
+      "step": 51540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611177967005347e-05,
+      "loss": 1.0811,
+      "step": 51550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611101802062547e-05,
+      "loss": 1.1015,
+      "step": 51560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.611025637119747e-05,
+      "loss": 1.031,
+      "step": 51570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6109494721769466e-05,
+      "loss": 1.0508,
+      "step": 51580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6108733072341464e-05,
+      "loss": 1.1223,
+      "step": 51590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610797142291347e-05,
+      "loss": 1.2941,
+      "step": 51600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610720977348546e-05,
+      "loss": 1.2714,
+      "step": 51610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610644812405746e-05,
+      "loss": 1.1122,
+      "step": 51620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610568647462946e-05,
+      "loss": 1.1251,
+      "step": 51630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610492482520146e-05,
+      "loss": 1.0556,
+      "step": 51640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6104163175773455e-05,
+      "loss": 1.0139,
+      "step": 51650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610340152634546e-05,
+      "loss": 1.1377,
+      "step": 51660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610263987691745e-05,
+      "loss": 0.9738,
+      "step": 51670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6101878227489454e-05,
+      "loss": 0.9891,
+      "step": 51680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610111657806145e-05,
+      "loss": 1.1966,
+      "step": 51690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.610035492863345e-05,
+      "loss": 1.0813,
+      "step": 51700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609959327920545e-05,
+      "loss": 1.099,
+      "step": 51710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609883162977745e-05,
+      "loss": 1.0352,
+      "step": 51720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609806998034945e-05,
+      "loss": 1.0325,
+      "step": 51730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6097308330921446e-05,
+      "loss": 1.2164,
+      "step": 51740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6096546681493443e-05,
+      "loss": 1.1243,
+      "step": 51750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609578503206544e-05,
+      "loss": 0.9728,
+      "step": 51760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6095023382637445e-05,
+      "loss": 1.0751,
+      "step": 51770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609426173320944e-05,
+      "loss": 1.1262,
+      "step": 51780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609350008378144e-05,
+      "loss": 1.0694,
+      "step": 51790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.609273843435344e-05,
+      "loss": 1.2254,
+      "step": 51800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.609197678492544e-05,
+      "loss": 1.0566,
+      "step": 51810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.609121513549743e-05,
+      "loss": 1.0063,
+      "step": 51820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.609045348606944e-05,
+      "loss": 1.1254,
+      "step": 51830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6089691836641434e-05,
+      "loss": 1.0723,
+      "step": 51840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608893018721343e-05,
+      "loss": 1.0105,
+      "step": 51850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608816853778543e-05,
+      "loss": 1.0179,
+      "step": 51860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608740688835743e-05,
+      "loss": 1.0039,
+      "step": 51870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6086645238929424e-05,
+      "loss": 1.2407,
+      "step": 51880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608588358950143e-05,
+      "loss": 1.0789,
+      "step": 51890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6085121940073426e-05,
+      "loss": 1.0856,
+      "step": 51900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608436029064542e-05,
+      "loss": 1.0224,
+      "step": 51910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608359864121742e-05,
+      "loss": 1.1542,
+      "step": 51920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608283699178942e-05,
+      "loss": 0.9465,
+      "step": 51930
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608207534236142e-05,
+      "loss": 1.1623,
+      "step": 51940
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608131369293342e-05,
+      "loss": 1.0365,
+      "step": 51950
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.608055204350542e-05,
+      "loss": 0.9647,
+      "step": 51960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6079790394077415e-05,
+      "loss": 1.2004,
+      "step": 51970
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.607902874464942e-05,
+      "loss": 1.2549,
+      "step": 51980
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.607826709522141e-05,
+      "loss": 1.1077,
+      "step": 51990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6077505445793414e-05,
+      "loss": 1.1532,
+      "step": 52000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.607674379636541e-05,
+      "loss": 1.1231,
+      "step": 52010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.607598214693741e-05,
+      "loss": 1.0013,
+      "step": 52020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6075220497509406e-05,
+      "loss": 1.1816,
+      "step": 52030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.607445884808141e-05,
+      "loss": 1.0788,
+      "step": 52040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60736971986534e-05,
+      "loss": 1.1128,
+      "step": 52050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6072935549225406e-05,
+      "loss": 1.2341,
+      "step": 52060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60721738997974e-05,
+      "loss": 1.0819,
+      "step": 52070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60714122503694e-05,
+      "loss": 1.0609,
+      "step": 52080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60706506009414e-05,
+      "loss": 1.1801,
+      "step": 52090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60698889515134e-05,
+      "loss": 1.1728,
+      "step": 52100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60691273020854e-05,
+      "loss": 1.0247,
+      "step": 52110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60683656526574e-05,
+      "loss": 1.0161,
+      "step": 52120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6067604003229395e-05,
+      "loss": 1.1442,
+      "step": 52130
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.606684235380139e-05,
+      "loss": 1.1254,
+      "step": 52140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6066080704373396e-05,
+      "loss": 1.1463,
+      "step": 52150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6065319054945394e-05,
+      "loss": 1.0711,
+      "step": 52160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.606455740551739e-05,
+      "loss": 1.1692,
+      "step": 52170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.606379575608939e-05,
+      "loss": 1.1534,
+      "step": 52180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.606303410666139e-05,
+      "loss": 1.2448,
+      "step": 52190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6062272457233384e-05,
+      "loss": 1.1942,
+      "step": 52200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.606151080780539e-05,
+      "loss": 1.1814,
+      "step": 52210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6060749158377385e-05,
+      "loss": 1.0568,
+      "step": 52220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.605998750894938e-05,
+      "loss": 1.1337,
+      "step": 52230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605922585952138e-05,
+      "loss": 1.2043,
+      "step": 52240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6058464210093384e-05,
+      "loss": 1.0381,
+      "step": 52250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6057702560665375e-05,
+      "loss": 1.0022,
+      "step": 52260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605694091123738e-05,
+      "loss": 1.1436,
+      "step": 52270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605617926180938e-05,
+      "loss": 1.0567,
+      "step": 52280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6055417612381374e-05,
+      "loss": 0.9287,
+      "step": 52290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605465596295337e-05,
+      "loss": 1.0182,
+      "step": 52300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6053894313525376e-05,
+      "loss": 0.9616,
+      "step": 52310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6053132664097374e-05,
+      "loss": 1.2662,
+      "step": 52320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605237101466937e-05,
+      "loss": 1.1966,
+      "step": 52330
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605160936524137e-05,
+      "loss": 1.1766,
+      "step": 52340
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6050847715813366e-05,
+      "loss": 1.0688,
+      "step": 52350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.605008606638537e-05,
+      "loss": 1.0485,
+      "step": 52360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604932441695736e-05,
+      "loss": 1.2001,
+      "step": 52370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6048562767529365e-05,
+      "loss": 1.1622,
+      "step": 52380
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604780111810136e-05,
+      "loss": 1.0916,
+      "step": 52390
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604703946867336e-05,
+      "loss": 1.2591,
+      "step": 52400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604627781924536e-05,
+      "loss": 1.1825,
+      "step": 52410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604551616981736e-05,
+      "loss": 1.1089,
+      "step": 52420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604475452038935e-05,
+      "loss": 1.1017,
+      "step": 52430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.604399287096136e-05,
+      "loss": 1.1726,
+      "step": 52440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6043231221533354e-05,
+      "loss": 1.0311,
+      "step": 52450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.604246957210535e-05,
+      "loss": 0.9803,
+      "step": 52460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.604170792267735e-05,
+      "loss": 1.2288,
+      "step": 52470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.604094627324935e-05,
+      "loss": 1.1298,
+      "step": 52480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.604018462382135e-05,
+      "loss": 0.9119,
+      "step": 52490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603942297439335e-05,
+      "loss": 1.0498,
+      "step": 52500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6038661324965346e-05,
+      "loss": 1.2318,
+      "step": 52510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603789967553734e-05,
+      "loss": 1.0923,
+      "step": 52520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603713802610935e-05,
+      "loss": 1.0332,
+      "step": 52530
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6036376376681345e-05,
+      "loss": 1.0411,
+      "step": 52540
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603561472725334e-05,
+      "loss": 1.1806,
+      "step": 52550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603485307782534e-05,
+      "loss": 1.1381,
+      "step": 52560
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6034091428397344e-05,
+      "loss": 1.0277,
+      "step": 52570
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6033329778969335e-05,
+      "loss": 1.06,
+      "step": 52580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603256812954134e-05,
+      "loss": 1.087,
+      "step": 52590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6031806480113336e-05,
+      "loss": 1.1521,
+      "step": 52600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6031044830685334e-05,
+      "loss": 1.1683,
+      "step": 52610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.603028318125733e-05,
+      "loss": 1.0817,
+      "step": 52620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6029521531829336e-05,
+      "loss": 1.006,
+      "step": 52630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6028759882401326e-05,
+      "loss": 1.065,
+      "step": 52640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.602799823297333e-05,
+      "loss": 0.9944,
+      "step": 52650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.602723658354533e-05,
+      "loss": 1.2183,
+      "step": 52660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6026474934117326e-05,
+      "loss": 1.0185,
+      "step": 52670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602571328468932e-05,
+      "loss": 0.9889,
+      "step": 52680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602495163526133e-05,
+      "loss": 1.1982,
+      "step": 52690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6024189985833325e-05,
+      "loss": 0.9669,
+      "step": 52700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602342833640532e-05,
+      "loss": 1.0929,
+      "step": 52710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602266668697732e-05,
+      "loss": 1.0311,
+      "step": 52720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602190503754932e-05,
+      "loss": 1.1995,
+      "step": 52730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602114338812132e-05,
+      "loss": 1.1011,
+      "step": 52740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.602038173869332e-05,
+      "loss": 1.0728,
+      "step": 52750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6019620089265316e-05,
+      "loss": 1.0825,
+      "step": 52760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6018858439837314e-05,
+      "loss": 1.212,
+      "step": 52770
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.601809679040932e-05,
+      "loss": 1.0582,
+      "step": 52780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.601733514098131e-05,
+      "loss": 1.0449,
+      "step": 52790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.601657349155331e-05,
+      "loss": 1.1535,
+      "step": 52800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.601581184212531e-05,
+      "loss": 1.0298,
+      "step": 52810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.601505019269731e-05,
+      "loss": 1.1217,
+      "step": 52820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6014288543269305e-05,
+      "loss": 1.0408,
+      "step": 52830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.60135268938413e-05,
+      "loss": 1.0741,
+      "step": 52840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.60127652444133e-05,
+      "loss": 1.2656,
+      "step": 52850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6012003594985305e-05,
+      "loss": 1.0138,
+      "step": 52860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.6011241945557295e-05,
+      "loss": 1.0526,
+      "step": 52870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.60104802961293e-05,
+      "loss": 1.1196,
+      "step": 52880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.60097186467013e-05,
+      "loss": 1.0135,
+      "step": 52890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.6008956997273294e-05,
+      "loss": 1.097,
+      "step": 52900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.60081953478453e-05,
+      "loss": 0.9981,
+      "step": 52910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.6007433698417296e-05,
+      "loss": 1.0836,
+      "step": 52920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.6006672048989294e-05,
+      "loss": 1.0712,
+      "step": 52930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.600591039956129e-05,
+      "loss": 1.1435,
+      "step": 52940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.6005148750133295e-05,
+      "loss": 1.0326,
+      "step": 52950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.6004387100705286e-05,
+      "loss": 1.223,
+      "step": 52960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.600362545127729e-05,
+      "loss": 1.0293,
+      "step": 52970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.600286380184929e-05,
+      "loss": 1.0469,
+      "step": 52980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.6002102152421285e-05,
+      "loss": 1.1228,
+      "step": 52990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.600134050299328e-05,
+      "loss": 0.9705,
+      "step": 53000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.600057885356529e-05,
+      "loss": 1.0137,
+      "step": 53010
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599981720413728e-05,
+      "loss": 1.0941,
+      "step": 53020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599905555470928e-05,
+      "loss": 0.9962,
+      "step": 53030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599829390528128e-05,
+      "loss": 1.0343,
+      "step": 53040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599753225585328e-05,
+      "loss": 1.2297,
+      "step": 53050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5996770606425274e-05,
+      "loss": 1.0575,
+      "step": 53060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599600895699728e-05,
+      "loss": 1.1268,
+      "step": 53070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599524730756927e-05,
+      "loss": 1.0468,
+      "step": 53080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599448565814127e-05,
+      "loss": 1.1415,
+      "step": 53090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.599372400871327e-05,
+      "loss": 1.1606,
+      "step": 53100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.599296235928527e-05,
+      "loss": 1.0706,
+      "step": 53110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.599220070985727e-05,
+      "loss": 1.1971,
+      "step": 53120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.599143906042927e-05,
+      "loss": 1.0309,
+      "step": 53130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.599067741100127e-05,
+      "loss": 1.0716,
+      "step": 53140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5989915761573265e-05,
+      "loss": 1.0015,
+      "step": 53150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598915411214527e-05,
+      "loss": 1.2188,
+      "step": 53160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598839246271726e-05,
+      "loss": 1.228,
+      "step": 53170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5987630813289264e-05,
+      "loss": 1.228,
+      "step": 53180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598686916386126e-05,
+      "loss": 0.9622,
+      "step": 53190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598610751443326e-05,
+      "loss": 1.2072,
+      "step": 53200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5985345865005257e-05,
+      "loss": 1.1015,
+      "step": 53210
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598458421557726e-05,
+      "loss": 1.0336,
+      "step": 53220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598382256614925e-05,
+      "loss": 1.1564,
+      "step": 53230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5983060916721256e-05,
+      "loss": 1.0944,
+      "step": 53240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598229926729325e-05,
+      "loss": 0.9947,
+      "step": 53250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598153761786525e-05,
+      "loss": 1.0512,
+      "step": 53260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598077596843725e-05,
+      "loss": 1.0393,
+      "step": 53270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.598001431900925e-05,
+      "loss": 1.0166,
+      "step": 53280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.597925266958125e-05,
+      "loss": 1.0598,
+      "step": 53290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.597849102015325e-05,
+      "loss": 1.1244,
+      "step": 53300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5977729370725245e-05,
+      "loss": 0.9924,
+      "step": 53310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.597696772129724e-05,
+      "loss": 1.0706,
+      "step": 53320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5976206071869246e-05,
+      "loss": 1.0406,
+      "step": 53330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.597544442244124e-05,
+      "loss": 1.2329,
+      "step": 53340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.597468277301324e-05,
+      "loss": 1.0322,
+      "step": 53350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.597392112358524e-05,
+      "loss": 1.1359,
+      "step": 53360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5973159474157236e-05,
+      "loss": 1.0505,
+      "step": 53370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5972397824729234e-05,
+      "loss": 1.0274,
+      "step": 53380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.597163617530124e-05,
+      "loss": 1.1364,
+      "step": 53390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.597087452587323e-05,
+      "loss": 1.1177,
+      "step": 53400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.597011287644523e-05,
+      "loss": 1.0357,
+      "step": 53410
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596935122701723e-05,
+      "loss": 1.1471,
+      "step": 53420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596858957758923e-05,
+      "loss": 1.0272,
+      "step": 53430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5967827928161225e-05,
+      "loss": 1.0417,
+      "step": 53440
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596706627873323e-05,
+      "loss": 0.9696,
+      "step": 53450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596630462930522e-05,
+      "loss": 1.182,
+      "step": 53460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5965542979877225e-05,
+      "loss": 1.1012,
+      "step": 53470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596478133044922e-05,
+      "loss": 1.0783,
+      "step": 53480
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596401968102122e-05,
+      "loss": 1.0703,
+      "step": 53490
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5963258031593224e-05,
+      "loss": 1.2335,
+      "step": 53500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596249638216522e-05,
+      "loss": 1.0961,
+      "step": 53510
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596173473273722e-05,
+      "loss": 1.1462,
+      "step": 53520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5960973083309216e-05,
+      "loss": 1.1291,
+      "step": 53530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.596021143388122e-05,
+      "loss": 1.1035,
+      "step": 53540
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.595944978445321e-05,
+      "loss": 0.9932,
+      "step": 53550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5958688135025215e-05,
+      "loss": 1.1589,
+      "step": 53560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.595792648559721e-05,
+      "loss": 1.0856,
+      "step": 53570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.595716483616921e-05,
+      "loss": 1.1271,
+      "step": 53580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.595640318674121e-05,
+      "loss": 1.0046,
+      "step": 53590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.595564153731321e-05,
+      "loss": 1.0462,
+      "step": 53600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59548798878852e-05,
+      "loss": 1.083,
+      "step": 53610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.595411823845721e-05,
+      "loss": 1.0982,
+      "step": 53620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5953356589029204e-05,
+      "loss": 1.0461,
+      "step": 53630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59525949396012e-05,
+      "loss": 1.0635,
+      "step": 53640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59518332901732e-05,
+      "loss": 1.0509,
+      "step": 53650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5951071640745203e-05,
+      "loss": 1.1132,
+      "step": 53660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5950309991317194e-05,
+      "loss": 1.1929,
+      "step": 53670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59495483418892e-05,
+      "loss": 1.0613,
+      "step": 53680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5948786692461196e-05,
+      "loss": 1.1148,
+      "step": 53690
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.594802504303319e-05,
+      "loss": 0.9854,
+      "step": 53700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59472633936052e-05,
+      "loss": 1.1708,
+      "step": 53710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5946501744177195e-05,
+      "loss": 1.1837,
+      "step": 53720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.594574009474919e-05,
+      "loss": 1.2412,
+      "step": 53730
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.594497844532119e-05,
+      "loss": 1.1161,
+      "step": 53740
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.594421679589319e-05,
+      "loss": 1.1055,
+      "step": 53750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5943455146465185e-05,
+      "loss": 1.1198,
+      "step": 53760
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.594269349703719e-05,
+      "loss": 1.173,
+      "step": 53770
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.594193184760918e-05,
+      "loss": 1.1279,
+      "step": 53780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5941170198181184e-05,
+      "loss": 1.1409,
+      "step": 53790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.594040854875318e-05,
+      "loss": 1.2115,
+      "step": 53800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593964689932518e-05,
+      "loss": 1.0619,
+      "step": 53810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5938885249897177e-05,
+      "loss": 1.0609,
+      "step": 53820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593812360046918e-05,
+      "loss": 1.1434,
+      "step": 53830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593736195104117e-05,
+      "loss": 1.0874,
+      "step": 53840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5936600301613176e-05,
+      "loss": 1.1083,
+      "step": 53850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593583865218517e-05,
+      "loss": 1.0292,
+      "step": 53860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593507700275717e-05,
+      "loss": 1.0963,
+      "step": 53870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593431535332917e-05,
+      "loss": 1.2726,
+      "step": 53880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593355370390117e-05,
+      "loss": 1.0237,
+      "step": 53890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593279205447317e-05,
+      "loss": 1.1646,
+      "step": 53900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593203040504517e-05,
+      "loss": 1.1053,
+      "step": 53910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593126875561717e-05,
+      "loss": 1.0015,
+      "step": 53920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.593050710618916e-05,
+      "loss": 1.1418,
+      "step": 53930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5929745456761166e-05,
+      "loss": 1.2713,
+      "step": 53940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5928983807333164e-05,
+      "loss": 1.014,
+      "step": 53950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.592822215790516e-05,
+      "loss": 1.0449,
+      "step": 53960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.592746050847716e-05,
+      "loss": 1.0264,
+      "step": 53970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.592669885904916e-05,
+      "loss": 1.0298,
+      "step": 53980
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5925937209621154e-05,
+      "loss": 0.9763,
+      "step": 53990
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.592517556019316e-05,
+      "loss": 1.0443,
+      "step": 54000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5924413910765155e-05,
+      "loss": 1.0594,
+      "step": 54010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.592365226133715e-05,
+      "loss": 1.0913,
+      "step": 54020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.592289061190915e-05,
+      "loss": 1.0571,
+      "step": 54030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5922128962481155e-05,
+      "loss": 1.2479,
+      "step": 54040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5921367313053145e-05,
+      "loss": 1.0173,
+      "step": 54050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.592060566362515e-05,
+      "loss": 1.1408,
+      "step": 54060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591984401419715e-05,
+      "loss": 0.995,
+      "step": 54070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5919082364769145e-05,
+      "loss": 1.1903,
+      "step": 54080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591832071534115e-05,
+      "loss": 1.1713,
+      "step": 54090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5917559065913146e-05,
+      "loss": 1.0754,
+      "step": 54100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5916797416485144e-05,
+      "loss": 1.1224,
+      "step": 54110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591603576705714e-05,
+      "loss": 0.9856,
+      "step": 54120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5915274117629145e-05,
+      "loss": 0.9863,
+      "step": 54130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5914512468201136e-05,
+      "loss": 1.2068,
+      "step": 54140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591375081877314e-05,
+      "loss": 1.0559,
+      "step": 54150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591298916934514e-05,
+      "loss": 1.103,
+      "step": 54160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5912227519917135e-05,
+      "loss": 1.1378,
+      "step": 54170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591146587048913e-05,
+      "loss": 1.0294,
+      "step": 54180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.591070422106114e-05,
+      "loss": 1.1678,
+      "step": 54190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.590994257163313e-05,
+      "loss": 1.0135,
+      "step": 54200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590918092220513e-05,
+      "loss": 1.0767,
+      "step": 54210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590841927277712e-05,
+      "loss": 1.065,
+      "step": 54220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590765762334913e-05,
+      "loss": 1.1528,
+      "step": 54230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5906895973921124e-05,
+      "loss": 1.0757,
+      "step": 54240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590613432449312e-05,
+      "loss": 1.1575,
+      "step": 54250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590537267506512e-05,
+      "loss": 1.0572,
+      "step": 54260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5904611025637124e-05,
+      "loss": 1.0252,
+      "step": 54270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590384937620912e-05,
+      "loss": 1.1742,
+      "step": 54280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590308772678112e-05,
+      "loss": 1.2325,
+      "step": 54290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590232607735312e-05,
+      "loss": 0.9576,
+      "step": 54300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590156442792511e-05,
+      "loss": 1.0362,
+      "step": 54310
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.590080277849712e-05,
+      "loss": 0.9329,
+      "step": 54320
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5900041129069115e-05,
+      "loss": 1.1295,
+      "step": 54330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.589927947964111e-05,
+      "loss": 1.1284,
+      "step": 54340
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.589851783021311e-05,
+      "loss": 1.2104,
+      "step": 54350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5897756180785114e-05,
+      "loss": 1.1086,
+      "step": 54360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5896994531357105e-05,
+      "loss": 1.131,
+      "step": 54370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.589623288192911e-05,
+      "loss": 1.0886,
+      "step": 54380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.589547123250111e-05,
+      "loss": 1.0576,
+      "step": 54390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5894709583073104e-05,
+      "loss": 1.0918,
+      "step": 54400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.58939479336451e-05,
+      "loss": 1.1569,
+      "step": 54410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5893186284217106e-05,
+      "loss": 1.0646,
+      "step": 54420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5892424634789097e-05,
+      "loss": 1.1033,
+      "step": 54430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58916629853611e-05,
+      "loss": 1.1074,
+      "step": 54440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58909013359331e-05,
+      "loss": 1.1391,
+      "step": 54450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5890139686505096e-05,
+      "loss": 1.1365,
+      "step": 54460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588937803707709e-05,
+      "loss": 1.1129,
+      "step": 54470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58886163876491e-05,
+      "loss": 1.1781,
+      "step": 54480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5887854738221095e-05,
+      "loss": 1.0435,
+      "step": 54490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588709308879309e-05,
+      "loss": 1.0212,
+      "step": 54500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5886331439365097e-05,
+      "loss": 1.0702,
+      "step": 54510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588556978993709e-05,
+      "loss": 1.0919,
+      "step": 54520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588480814050909e-05,
+      "loss": 1.0671,
+      "step": 54530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588404649108109e-05,
+      "loss": 1.0552,
+      "step": 54540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5883284841653086e-05,
+      "loss": 1.0462,
+      "step": 54550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5882523192225084e-05,
+      "loss": 1.1449,
+      "step": 54560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588176154279709e-05,
+      "loss": 1.0172,
+      "step": 54570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588099989336908e-05,
+      "loss": 1.0181,
+      "step": 54580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.588023824394108e-05,
+      "loss": 1.1597,
+      "step": 54590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.587947659451308e-05,
+      "loss": 1.0252,
+      "step": 54600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.587871494508508e-05,
+      "loss": 1.2456,
+      "step": 54610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5877953295657075e-05,
+      "loss": 1.0123,
+      "step": 54620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.587719164622908e-05,
+      "loss": 1.0552,
+      "step": 54630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.587642999680107e-05,
+      "loss": 1.0733,
+      "step": 54640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5875668347373075e-05,
+      "loss": 1.0923,
+      "step": 54650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.587490669794507e-05,
+      "loss": 1.021,
+      "step": 54660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.587414504851707e-05,
+      "loss": 1.1435,
+      "step": 54670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.587338339908907e-05,
+      "loss": 0.9979,
+      "step": 54680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5872621749661065e-05,
+      "loss": 1.2307,
+      "step": 54690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.587186010023307e-05,
+      "loss": 1.255,
+      "step": 54700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5871098450805066e-05,
+      "loss": 1.1678,
+      "step": 54710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5870336801377064e-05,
+      "loss": 1.0377,
+      "step": 54720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586957515194906e-05,
+      "loss": 1.0799,
+      "step": 54730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5868813502521065e-05,
+      "loss": 1.1317,
+      "step": 54740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5868051853093056e-05,
+      "loss": 1.1075,
+      "step": 54750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586729020366506e-05,
+      "loss": 1.1092,
+      "step": 54760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586652855423706e-05,
+      "loss": 1.1482,
+      "step": 54770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5865766904809055e-05,
+      "loss": 0.9897,
+      "step": 54780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586500525538105e-05,
+      "loss": 1.0166,
+      "step": 54790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586424360595306e-05,
+      "loss": 1.1384,
+      "step": 54800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586348195652505e-05,
+      "loss": 1.1164,
+      "step": 54810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586272030709705e-05,
+      "loss": 1.1889,
+      "step": 54820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586195865766905e-05,
+      "loss": 1.0653,
+      "step": 54830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.586119700824105e-05,
+      "loss": 1.1942,
+      "step": 54840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5860435358813044e-05,
+      "loss": 1.1216,
+      "step": 54850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585967370938505e-05,
+      "loss": 1.0246,
+      "step": 54860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5858912059957046e-05,
+      "loss": 1.1217,
+      "step": 54870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5858150410529044e-05,
+      "loss": 1.0118,
+      "step": 54880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585738876110105e-05,
+      "loss": 1.1645,
+      "step": 54890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585662711167304e-05,
+      "loss": 1.0869,
+      "step": 54900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585586546224504e-05,
+      "loss": 1.0422,
+      "step": 54910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585510381281704e-05,
+      "loss": 1.047,
+      "step": 54920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585434216338904e-05,
+      "loss": 1.0953,
+      "step": 54930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5853580513961035e-05,
+      "loss": 1.1706,
+      "step": 54940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585281886453304e-05,
+      "loss": 1.0565,
+      "step": 54950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585205721510503e-05,
+      "loss": 1.1137,
+      "step": 54960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5851295565677034e-05,
+      "loss": 1.1566,
+      "step": 54970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.585053391624903e-05,
+      "loss": 1.188,
+      "step": 54980
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584977226682103e-05,
+      "loss": 1.1115,
+      "step": 54990
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584901061739303e-05,
+      "loss": 1.0047,
+      "step": 55000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584824896796503e-05,
+      "loss": 1.1188,
+      "step": 55010
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584748731853702e-05,
+      "loss": 0.996,
+      "step": 55020
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5846725669109026e-05,
+      "loss": 0.9542,
+      "step": 55030
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584596401968102e-05,
+      "loss": 1.0479,
+      "step": 55040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584520237025302e-05,
+      "loss": 1.0015,
+      "step": 55050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584444072082502e-05,
+      "loss": 1.1857,
+      "step": 55060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.584367907139702e-05,
+      "loss": 1.0047,
+      "step": 55070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.584291742196902e-05,
+      "loss": 1.0476,
+      "step": 55080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.584215577254102e-05,
+      "loss": 1.0247,
+      "step": 55090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.584139412311302e-05,
+      "loss": 1.0059,
+      "step": 55100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.584063247368501e-05,
+      "loss": 1.0991,
+      "step": 55110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5839870824257017e-05,
+      "loss": 1.1979,
+      "step": 55120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.583910917482901e-05,
+      "loss": 1.1072,
+      "step": 55130
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.583834752540101e-05,
+      "loss": 1.1082,
+      "step": 55140
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.583758587597301e-05,
+      "loss": 1.017,
+      "step": 55150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5836824226545006e-05,
+      "loss": 0.9612,
+      "step": 55160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5836062577117004e-05,
+      "loss": 1.017,
+      "step": 55170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.583530092768901e-05,
+      "loss": 1.0416,
+      "step": 55180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5834539278261e-05,
+      "loss": 1.1476,
+      "step": 55190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5833777628833e-05,
+      "loss": 1.125,
+      "step": 55200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5833015979405e-05,
+      "loss": 1.083,
+      "step": 55210
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5832254329977e-05,
+      "loss": 1.0689,
+      "step": 55220
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5831492680548996e-05,
+      "loss": 1.0891,
+      "step": 55230
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5830731031121e-05,
+      "loss": 1.2617,
+      "step": 55240
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5829969381693e-05,
+      "loss": 1.0164,
+      "step": 55250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5829207732264995e-05,
+      "loss": 0.9704,
+      "step": 55260
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.582844608283699e-05,
+      "loss": 1.1252,
+      "step": 55270
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.582768443340899e-05,
+      "loss": 1.3583,
+      "step": 55280
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5826922783980994e-05,
+      "loss": 1.1535,
+      "step": 55290
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582616113455299e-05,
+      "loss": 1.1533,
+      "step": 55300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582539948512499e-05,
+      "loss": 1.0717,
+      "step": 55310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5824637835696986e-05,
+      "loss": 1.0061,
+      "step": 55320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582387618626899e-05,
+      "loss": 1.1006,
+      "step": 55330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582311453684098e-05,
+      "loss": 1.0122,
+      "step": 55340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5822352887412985e-05,
+      "loss": 1.1014,
+      "step": 55350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582159123798498e-05,
+      "loss": 1.1154,
+      "step": 55360
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582082958855698e-05,
+      "loss": 1.0808,
+      "step": 55370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.582006793912898e-05,
+      "loss": 1.0774,
+      "step": 55380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581930628970098e-05,
+      "loss": 1.2509,
+      "step": 55390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581854464027297e-05,
+      "loss": 1.1391,
+      "step": 55400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581778299084498e-05,
+      "loss": 0.997,
+      "step": 55410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5817021341416974e-05,
+      "loss": 1.346,
+      "step": 55420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581625969198897e-05,
+      "loss": 1.1909,
+      "step": 55430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581549804256097e-05,
+      "loss": 1.0566,
+      "step": 55440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5814736393132974e-05,
+      "loss": 1.0251,
+      "step": 55450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581397474370497e-05,
+      "loss": 1.0167,
+      "step": 55460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581321309427697e-05,
+      "loss": 1.1218,
+      "step": 55470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5812451444848966e-05,
+      "loss": 1.0307,
+      "step": 55480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5811689795420964e-05,
+      "loss": 1.1627,
+      "step": 55490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.581092814599297e-05,
+      "loss": 1.0721,
+      "step": 55500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5810166496564965e-05,
+      "loss": 0.9749,
+      "step": 55510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580940484713696e-05,
+      "loss": 1.0664,
+      "step": 55520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580864319770896e-05,
+      "loss": 1.0817,
+      "step": 55530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5807881548280964e-05,
+      "loss": 1.178,
+      "step": 55540
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5807119898852955e-05,
+      "loss": 1.1212,
+      "step": 55550
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580635824942496e-05,
+      "loss": 1.0362,
+      "step": 55560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580559659999696e-05,
+      "loss": 1.061,
+      "step": 55570
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5804834950568954e-05,
+      "loss": 1.1042,
+      "step": 55580
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580407330114095e-05,
+      "loss": 1.1271,
+      "step": 55590
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580331165171295e-05,
+      "loss": 1.1341,
+      "step": 55600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580255000228495e-05,
+      "loss": 1.0905,
+      "step": 55610
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580178835285695e-05,
+      "loss": 1.1127,
+      "step": 55620
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.580102670342895e-05,
+      "loss": 1.0462,
+      "step": 55630
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5800265054000946e-05,
+      "loss": 1.0879,
+      "step": 55640
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579950340457294e-05,
+      "loss": 0.9752,
+      "step": 55650
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579874175514494e-05,
+      "loss": 1.0989,
+      "step": 55660
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5797980105716945e-05,
+      "loss": 1.1098,
+      "step": 55670
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579721845628894e-05,
+      "loss": 1.1284,
+      "step": 55680
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579645680686094e-05,
+      "loss": 1.0419,
+      "step": 55690
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579569515743294e-05,
+      "loss": 1.0355,
+      "step": 55700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579493350800494e-05,
+      "loss": 1.0577,
+      "step": 55710
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.579417185857693e-05,
+      "loss": 1.0159,
+      "step": 55720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5793410209148937e-05,
+      "loss": 1.0853,
+      "step": 55730
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5792648559720934e-05,
+      "loss": 1.0358,
+      "step": 55740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.579188691029293e-05,
+      "loss": 1.0883,
+      "step": 55750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.579112526086493e-05,
+      "loss": 1.0477,
+      "step": 55760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.579036361143693e-05,
+      "loss": 1.1599,
+      "step": 55770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5789601962008924e-05,
+      "loss": 1.1004,
+      "step": 55780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578884031258093e-05,
+      "loss": 1.2035,
+      "step": 55790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5788078663152926e-05,
+      "loss": 1.039,
+      "step": 55800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578731701372492e-05,
+      "loss": 1.1532,
+      "step": 55810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578655536429692e-05,
+      "loss": 1.1097,
+      "step": 55820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5785793714868925e-05,
+      "loss": 1.0327,
+      "step": 55830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578503206544092e-05,
+      "loss": 1.266,
+      "step": 55840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578427041601292e-05,
+      "loss": 1.205,
+      "step": 55850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578350876658492e-05,
+      "loss": 1.0657,
+      "step": 55860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5782747117156915e-05,
+      "loss": 1.0648,
+      "step": 55870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.578198546772892e-05,
+      "loss": 1.0731,
+      "step": 55880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5781223818300916e-05,
+      "loss": 1.0551,
+      "step": 55890
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5780462168872914e-05,
+      "loss": 1.0497,
+      "step": 55900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.577970051944491e-05,
+      "loss": 1.0901,
+      "step": 55910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5778938870016916e-05,
+      "loss": 1.2062,
+      "step": 55920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5778177220588906e-05,
+      "loss": 0.9392,
+      "step": 55930
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.577741557116091e-05,
+      "loss": 1.1502,
+      "step": 55940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.577665392173291e-05,
+      "loss": 1.2632,
+      "step": 55950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5775892272304905e-05,
+      "loss": 0.9644,
+      "step": 55960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57751306228769e-05,
+      "loss": 1.11,
+      "step": 55970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.577436897344891e-05,
+      "loss": 1.1902,
+      "step": 55980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57736073240209e-05,
+      "loss": 1.1918,
+      "step": 55990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57728456745929e-05,
+      "loss": 1.1267,
+      "step": 56000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57720840251649e-05,
+      "loss": 0.993,
+      "step": 56010
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57713223757369e-05,
+      "loss": 1.0981,
+      "step": 56020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5770560726308894e-05,
+      "loss": 1.2068,
+      "step": 56030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57697990768809e-05,
+      "loss": 1.0507,
+      "step": 56040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5769037427452896e-05,
+      "loss": 1.0588,
+      "step": 56050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5768275778024894e-05,
+      "loss": 1.1066,
+      "step": 56060
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.576751412859689e-05,
+      "loss": 1.1373,
+      "step": 56070
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.576675247916889e-05,
+      "loss": 1.0582,
+      "step": 56080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.576599082974089e-05,
+      "loss": 1.1991,
+      "step": 56090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5765229180312884e-05,
+      "loss": 1.1523,
+      "step": 56100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.576446753088489e-05,
+      "loss": 1.047,
+      "step": 56110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5763705881456885e-05,
+      "loss": 1.0146,
+      "step": 56120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.576294423202888e-05,
+      "loss": 1.2485,
+      "step": 56130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.576218258260088e-05,
+      "loss": 0.9678,
+      "step": 56140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5761420933172884e-05,
+      "loss": 1.1977,
+      "step": 56150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5760659283744875e-05,
+      "loss": 0.9693,
+      "step": 56160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.575989763431688e-05,
+      "loss": 1.1177,
+      "step": 56170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575913598488888e-05,
+      "loss": 1.086,
+      "step": 56180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5758374335460874e-05,
+      "loss": 1.1037,
+      "step": 56190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575761268603287e-05,
+      "loss": 0.9762,
+      "step": 56200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5756851036604876e-05,
+      "loss": 0.9839,
+      "step": 56210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5756089387176873e-05,
+      "loss": 1.2385,
+      "step": 56220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575532773774887e-05,
+      "loss": 1.0282,
+      "step": 56230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575456608832087e-05,
+      "loss": 1.0642,
+      "step": 56240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5753804438892866e-05,
+      "loss": 1.2883,
+      "step": 56250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575304278946487e-05,
+      "loss": 1.2097,
+      "step": 56260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575228114003687e-05,
+      "loss": 1.074,
+      "step": 56270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5751519490608865e-05,
+      "loss": 1.0283,
+      "step": 56280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.575075784118086e-05,
+      "loss": 1.0939,
+      "step": 56290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574999619175287e-05,
+      "loss": 0.9863,
+      "step": 56300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574923454232486e-05,
+      "loss": 1.1782,
+      "step": 56310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574847289289686e-05,
+      "loss": 1.0842,
+      "step": 56320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574771124346886e-05,
+      "loss": 1.2767,
+      "step": 56330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574694959404086e-05,
+      "loss": 1.1006,
+      "step": 56340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5746187944612854e-05,
+      "loss": 1.0181,
+      "step": 56350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574542629518486e-05,
+      "loss": 1.0632,
+      "step": 56360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574466464575685e-05,
+      "loss": 1.0838,
+      "step": 56370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574390299632885e-05,
+      "loss": 1.2655,
+      "step": 56380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.574314134690085e-05,
+      "loss": 1.1137,
+      "step": 56390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.574237969747285e-05,
+      "loss": 1.2721,
+      "step": 56400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5741618048044846e-05,
+      "loss": 1.0337,
+      "step": 56410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.574085639861685e-05,
+      "loss": 1.0912,
+      "step": 56420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.574009474918885e-05,
+      "loss": 0.9985,
+      "step": 56430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5739333099760845e-05,
+      "loss": 1.0613,
+      "step": 56440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573857145033284e-05,
+      "loss": 1.1063,
+      "step": 56450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573780980090484e-05,
+      "loss": 1.1609,
+      "step": 56460
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5737048151476844e-05,
+      "loss": 1.1039,
+      "step": 56470
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573628650204884e-05,
+      "loss": 1.0866,
+      "step": 56480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573552485262084e-05,
+      "loss": 1.332,
+      "step": 56490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5734763203192836e-05,
+      "loss": 1.041,
+      "step": 56500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573400155376484e-05,
+      "loss": 1.0365,
+      "step": 56510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573323990433683e-05,
+      "loss": 1.2391,
+      "step": 56520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5732478254908836e-05,
+      "loss": 1.0041,
+      "step": 56530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5731716605480826e-05,
+      "loss": 1.2011,
+      "step": 56540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573095495605283e-05,
+      "loss": 1.061,
+      "step": 56550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.573019330662483e-05,
+      "loss": 1.0359,
+      "step": 56560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5729431657196825e-05,
+      "loss": 1.198,
+      "step": 56570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.572867000776882e-05,
+      "loss": 1.1283,
+      "step": 56580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.572790835834083e-05,
+      "loss": 1.0123,
+      "step": 56590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5727146708912825e-05,
+      "loss": 1.0378,
+      "step": 56600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.572638505948482e-05,
+      "loss": 1.0603,
+      "step": 56610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.572562341005682e-05,
+      "loss": 1.1282,
+      "step": 56620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.572486176062882e-05,
+      "loss": 1.1468,
+      "step": 56630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.572410011120082e-05,
+      "loss": 1.1882,
+      "step": 56640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.572333846177282e-05,
+      "loss": 1.0889,
+      "step": 56650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5722576812344816e-05,
+      "loss": 0.9852,
+      "step": 56660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5721815162916814e-05,
+      "loss": 1.1698,
+      "step": 56670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.572105351348882e-05,
+      "loss": 1.0856,
+      "step": 56680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.572029186406081e-05,
+      "loss": 1.2451,
+      "step": 56690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.571953021463281e-05,
+      "loss": 1.1915,
+      "step": 56700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.571876856520481e-05,
+      "loss": 1.0652,
+      "step": 56710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.571800691577681e-05,
+      "loss": 1.0415,
+      "step": 56720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5717245266348805e-05,
+      "loss": 1.179,
+      "step": 56730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.571648361692081e-05,
+      "loss": 1.0264,
+      "step": 56740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.57157219674928e-05,
+      "loss": 1.1055,
+      "step": 56750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5714960318064804e-05,
+      "loss": 1.1472,
+      "step": 56760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.57141986686368e-05,
+      "loss": 0.9315,
+      "step": 56770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.57134370192088e-05,
+      "loss": 1.1763,
+      "step": 56780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.57126753697808e-05,
+      "loss": 0.9398,
+      "step": 56790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.57119137203528e-05,
+      "loss": 1.0841,
+      "step": 56800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.57111520709248e-05,
+      "loss": 1.08,
+      "step": 56810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5710390421496796e-05,
+      "loss": 1.0519,
+      "step": 56820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5709628772068793e-05,
+      "loss": 1.219,
+      "step": 56830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570886712264079e-05,
+      "loss": 1.1372,
+      "step": 56840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5708105473212795e-05,
+      "loss": 1.1739,
+      "step": 56850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570734382378479e-05,
+      "loss": 1.1199,
+      "step": 56860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570658217435679e-05,
+      "loss": 1.1706,
+      "step": 56870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570582052492879e-05,
+      "loss": 1.0653,
+      "step": 56880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570505887550079e-05,
+      "loss": 1.0488,
+      "step": 56890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570429722607278e-05,
+      "loss": 1.0858,
+      "step": 56900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570353557664479e-05,
+      "loss": 1.0422,
+      "step": 56910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5702773927216784e-05,
+      "loss": 1.0981,
+      "step": 56920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570201227778878e-05,
+      "loss": 1.0094,
+      "step": 56930
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.570125062836078e-05,
+      "loss": 1.1506,
+      "step": 56940
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5700488978932783e-05,
+      "loss": 1.134,
+      "step": 56950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5699727329504774e-05,
+      "loss": 1.0794,
+      "step": 56960
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569896568007678e-05,
+      "loss": 1.0382,
+      "step": 56970
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5698204030648776e-05,
+      "loss": 1.0318,
+      "step": 56980
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569744238122077e-05,
+      "loss": 1.13,
+      "step": 56990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569668073179277e-05,
+      "loss": 1.0833,
+      "step": 57000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569591908236477e-05,
+      "loss": 1.2002,
+      "step": 57010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569515743293677e-05,
+      "loss": 1.0566,
+      "step": 57020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569439578350877e-05,
+      "loss": 1.0419,
+      "step": 57030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.569363413408077e-05,
+      "loss": 1.0639,
+      "step": 57040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5692872484652765e-05,
+      "loss": 1.1081,
+      "step": 57050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.569211083522477e-05,
+      "loss": 1.0885,
+      "step": 57060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.569134918579676e-05,
+      "loss": 1.1732,
+      "step": 57070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5690587536368764e-05,
+      "loss": 1.0864,
+      "step": 57080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568982588694076e-05,
+      "loss": 1.1907,
+      "step": 57090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568906423751276e-05,
+      "loss": 1.0072,
+      "step": 57100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5688302588084756e-05,
+      "loss": 1.2525,
+      "step": 57110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568754093865676e-05,
+      "loss": 1.1397,
+      "step": 57120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568677928922875e-05,
+      "loss": 0.9986,
+      "step": 57130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5686017639800756e-05,
+      "loss": 1.0604,
+      "step": 57140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568525599037275e-05,
+      "loss": 1.0767,
+      "step": 57150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568449434094475e-05,
+      "loss": 1.0452,
+      "step": 57160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568373269151675e-05,
+      "loss": 1.1941,
+      "step": 57170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568297104208875e-05,
+      "loss": 1.0688,
+      "step": 57180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568220939266075e-05,
+      "loss": 1.076,
+      "step": 57190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.568144774323275e-05,
+      "loss": 1.1405,
+      "step": 57200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5680686093804745e-05,
+      "loss": 1.1446,
+      "step": 57210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.567992444437674e-05,
+      "loss": 1.1014,
+      "step": 57220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5679162794948746e-05,
+      "loss": 1.1529,
+      "step": 57230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5678401145520744e-05,
+      "loss": 0.9562,
+      "step": 57240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.567763949609274e-05,
+      "loss": 1.0509,
+      "step": 57250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.567687784666474e-05,
+      "loss": 1.0869,
+      "step": 57260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.567611619723674e-05,
+      "loss": 1.1717,
+      "step": 57270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5675354547808734e-05,
+      "loss": 1.0572,
+      "step": 57280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.567459289838074e-05,
+      "loss": 1.1163,
+      "step": 57290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5673831248952735e-05,
+      "loss": 1.1689,
+      "step": 57300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.567306959952473e-05,
+      "loss": 0.9963,
+      "step": 57310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.567230795009673e-05,
+      "loss": 1.1188,
+      "step": 57320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5671546300668735e-05,
+      "loss": 1.0589,
+      "step": 57330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5670784651240725e-05,
+      "loss": 1.1015,
+      "step": 57340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.567002300181273e-05,
+      "loss": 0.9604,
+      "step": 57350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566926135238473e-05,
+      "loss": 1.081,
+      "step": 57360
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5668499702956724e-05,
+      "loss": 1.1211,
+      "step": 57370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566773805352872e-05,
+      "loss": 1.0231,
+      "step": 57380
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5666976404100726e-05,
+      "loss": 1.0455,
+      "step": 57390
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5666214754672724e-05,
+      "loss": 1.0487,
+      "step": 57400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566545310524472e-05,
+      "loss": 1.1,
+      "step": 57410
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566469145581672e-05,
+      "loss": 1.086,
+      "step": 57420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5663929806388716e-05,
+      "loss": 1.0482,
+      "step": 57430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566316815696072e-05,
+      "loss": 0.9929,
+      "step": 57440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566240650753271e-05,
+      "loss": 1.0219,
+      "step": 57450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5661644858104715e-05,
+      "loss": 1.1073,
+      "step": 57460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566088320867671e-05,
+      "loss": 1.1032,
+      "step": 57470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.566012155924871e-05,
+      "loss": 0.9811,
+      "step": 57480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.565935990982071e-05,
+      "loss": 1.1695,
+      "step": 57490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.565859826039271e-05,
+      "loss": 1.0589,
+      "step": 57500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56578366109647e-05,
+      "loss": 1.1435,
+      "step": 57510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.565707496153671e-05,
+      "loss": 1.0293,
+      "step": 57520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5656313312108704e-05,
+      "loss": 1.1124,
+      "step": 57530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56555516626807e-05,
+      "loss": 1.0507,
+      "step": 57540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56547900132527e-05,
+      "loss": 1.1059,
+      "step": 57550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5654028363824703e-05,
+      "loss": 1.1454,
+      "step": 57560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5653266714396694e-05,
+      "loss": 1.0354,
+      "step": 57570
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56525050649687e-05,
+      "loss": 1.0789,
+      "step": 57580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5651743415540696e-05,
+      "loss": 1.0932,
+      "step": 57590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.565098176611269e-05,
+      "loss": 1.1314,
+      "step": 57600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56502201166847e-05,
+      "loss": 1.1719,
+      "step": 57610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5649458467256695e-05,
+      "loss": 1.0047,
+      "step": 57620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.564869681782869e-05,
+      "loss": 1.1907,
+      "step": 57630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.564793516840069e-05,
+      "loss": 1.1731,
+      "step": 57640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5647173518972694e-05,
+      "loss": 1.1443,
+      "step": 57650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5646411869544685e-05,
+      "loss": 0.9896,
+      "step": 57660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.564565022011669e-05,
+      "loss": 1.0408,
+      "step": 57670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5644888570688687e-05,
+      "loss": 1.2683,
+      "step": 57680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5644126921260684e-05,
+      "loss": 1.1744,
+      "step": 57690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.564336527183268e-05,
+      "loss": 0.9836,
+      "step": 57700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5642603622404686e-05,
+      "loss": 1.0786,
+      "step": 57710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5641841972976676e-05,
+      "loss": 1.1603,
+      "step": 57720
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.564108032354868e-05,
+      "loss": 1.0779,
+      "step": 57730
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.564031867412068e-05,
+      "loss": 1.1167,
+      "step": 57740
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5639557024692676e-05,
+      "loss": 1.0802,
+      "step": 57750
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563879537526467e-05,
+      "loss": 1.0071,
+      "step": 57760
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563803372583668e-05,
+      "loss": 1.0235,
+      "step": 57770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5637272076408675e-05,
+      "loss": 1.0787,
+      "step": 57780
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563651042698067e-05,
+      "loss": 0.9728,
+      "step": 57790
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563574877755267e-05,
+      "loss": 1.0099,
+      "step": 57800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563498712812467e-05,
+      "loss": 1.2071,
+      "step": 57810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563422547869667e-05,
+      "loss": 1.0662,
+      "step": 57820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563346382926867e-05,
+      "loss": 0.9779,
+      "step": 57830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5632702179840666e-05,
+      "loss": 0.9435,
+      "step": 57840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5631940530412664e-05,
+      "loss": 1.0671,
+      "step": 57850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563117888098467e-05,
+      "loss": 1.0694,
+      "step": 57860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.563041723155666e-05,
+      "loss": 1.036,
+      "step": 57870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.562965558212866e-05,
+      "loss": 1.0388,
+      "step": 57880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.562889393270066e-05,
+      "loss": 0.9333,
+      "step": 57890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.562813228327266e-05,
+      "loss": 1.0661,
+      "step": 57900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5627370633844655e-05,
+      "loss": 1.066,
+      "step": 57910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.562660898441665e-05,
+      "loss": 1.1276,
+      "step": 57920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.562584733498865e-05,
+      "loss": 1.1668,
+      "step": 57930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5625085685560655e-05,
+      "loss": 1.0453,
+      "step": 57940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5624324036132645e-05,
+      "loss": 1.053,
+      "step": 57950
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.562356238670465e-05,
+      "loss": 1.1177,
+      "step": 57960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.562280073727665e-05,
+      "loss": 1.0497,
+      "step": 57970
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5622039087848644e-05,
+      "loss": 0.8769,
+      "step": 57980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.562127743842065e-05,
+      "loss": 1.0334,
+      "step": 57990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5620515788992646e-05,
+      "loss": 1.1522,
+      "step": 58000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5619754139564644e-05,
+      "loss": 1.1014,
+      "step": 58010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561899249013664e-05,
+      "loss": 1.2413,
+      "step": 58020
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5618230840708645e-05,
+      "loss": 1.1092,
+      "step": 58030
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5617469191280636e-05,
+      "loss": 1.1661,
+      "step": 58040
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561670754185264e-05,
+      "loss": 1.1059,
+      "step": 58050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561594589242464e-05,
+      "loss": 1.1443,
+      "step": 58060
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5615184242996635e-05,
+      "loss": 1.0004,
+      "step": 58070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561442259356863e-05,
+      "loss": 1.1823,
+      "step": 58080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561366094414064e-05,
+      "loss": 1.0212,
+      "step": 58090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561289929471263e-05,
+      "loss": 1.1234,
+      "step": 58100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561213764528463e-05,
+      "loss": 1.0807,
+      "step": 58110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561137599585663e-05,
+      "loss": 1.1803,
+      "step": 58120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.561061434642863e-05,
+      "loss": 1.1163,
+      "step": 58130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5609852697000624e-05,
+      "loss": 1.1683,
+      "step": 58140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560909104757263e-05,
+      "loss": 1.0914,
+      "step": 58150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560832939814462e-05,
+      "loss": 0.9974,
+      "step": 58160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5607567748716623e-05,
+      "loss": 1.0793,
+      "step": 58170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560680609928862e-05,
+      "loss": 1.1625,
+      "step": 58180
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560604444986062e-05,
+      "loss": 1.1853,
+      "step": 58190
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560528280043262e-05,
+      "loss": 1.1959,
+      "step": 58200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560452115100462e-05,
+      "loss": 1.1255,
+      "step": 58210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560375950157662e-05,
+      "loss": 1.1184,
+      "step": 58220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5602997852148615e-05,
+      "loss": 1.1013,
+      "step": 58230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560223620272062e-05,
+      "loss": 1.0463,
+      "step": 58240
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.560147455329261e-05,
+      "loss": 1.0511,
+      "step": 58250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5600712903864614e-05,
+      "loss": 0.9979,
+      "step": 58260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.559995125443661e-05,
+      "loss": 1.0725,
+      "step": 58270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.559918960500861e-05,
+      "loss": 1.0778,
+      "step": 58280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5598427955580607e-05,
+      "loss": 1.1707,
+      "step": 58290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.559766630615261e-05,
+      "loss": 1.0737,
+      "step": 58300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.55969046567246e-05,
+      "loss": 1.1449,
+      "step": 58310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5596143007296606e-05,
+      "loss": 1.0267,
+      "step": 58320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.55953813578686e-05,
+      "loss": 1.1089,
+      "step": 58330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.55946197084406e-05,
+      "loss": 0.9521,
+      "step": 58340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.55938580590126e-05,
+      "loss": 1.1339,
+      "step": 58350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5593096409584596e-05,
+      "loss": 0.9946,
+      "step": 58360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.559233476015659e-05,
+      "loss": 1.0713,
+      "step": 58370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.55915731107286e-05,
+      "loss": 1.0261,
+      "step": 58380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5590811461300595e-05,
+      "loss": 1.2841,
+      "step": 58390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.559004981187259e-05,
+      "loss": 1.1966,
+      "step": 58400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5589288162444597e-05,
+      "loss": 1.0653,
+      "step": 58410
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558852651301659e-05,
+      "loss": 0.9791,
+      "step": 58420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558776486358859e-05,
+      "loss": 1.1389,
+      "step": 58430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558700321416059e-05,
+      "loss": 1.0741,
+      "step": 58440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5586241564732586e-05,
+      "loss": 1.1268,
+      "step": 58450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5585479915304584e-05,
+      "loss": 1.0582,
+      "step": 58460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558471826587659e-05,
+      "loss": 1.0362,
+      "step": 58470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558395661644858e-05,
+      "loss": 1.1864,
+      "step": 58480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558319496702058e-05,
+      "loss": 1.0308,
+      "step": 58490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558243331759258e-05,
+      "loss": 1.1282,
+      "step": 58500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558167166816458e-05,
+      "loss": 1.0886,
+      "step": 58510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5580910018736575e-05,
+      "loss": 1.0581,
+      "step": 58520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.558014836930858e-05,
+      "loss": 1.0715,
+      "step": 58530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.557938671988057e-05,
+      "loss": 1.1056,
+      "step": 58540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5578625070452575e-05,
+      "loss": 1.0523,
+      "step": 58550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.557786342102457e-05,
+      "loss": 1.1976,
+      "step": 58560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.557710177159657e-05,
+      "loss": 1.0606,
+      "step": 58570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5576340122168574e-05,
+      "loss": 1.0933,
+      "step": 58580
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.557557847274057e-05,
+      "loss": 1.1045,
+      "step": 58590
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.557481682331257e-05,
+      "loss": 1.1899,
+      "step": 58600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5574055173884566e-05,
+      "loss": 0.9358,
+      "step": 58610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.557329352445657e-05,
+      "loss": 1.0067,
+      "step": 58620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.557253187502856e-05,
+      "loss": 1.0568,
+      "step": 58630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5571770225600565e-05,
+      "loss": 1.0961,
+      "step": 58640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.557100857617256e-05,
+      "loss": 1.1145,
+      "step": 58650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.557024692674456e-05,
+      "loss": 1.0013,
+      "step": 58660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556948527731656e-05,
+      "loss": 1.0856,
+      "step": 58670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556872362788856e-05,
+      "loss": 1.1799,
+      "step": 58680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556796197846055e-05,
+      "loss": 1.0355,
+      "step": 58690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556720032903256e-05,
+      "loss": 1.1133,
+      "step": 58700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5566438679604554e-05,
+      "loss": 1.1013,
+      "step": 58710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556567703017655e-05,
+      "loss": 1.0434,
+      "step": 58720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556491538074855e-05,
+      "loss": 1.0034,
+      "step": 58730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5564153731320554e-05,
+      "loss": 1.1479,
+      "step": 58740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5563392081892544e-05,
+      "loss": 1.2116,
+      "step": 58750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556263043246455e-05,
+      "loss": 1.0509,
+      "step": 58760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5561868783036546e-05,
+      "loss": 1.0489,
+      "step": 58770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5561107133608543e-05,
+      "loss": 1.0028,
+      "step": 58780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.556034548418055e-05,
+      "loss": 1.0298,
+      "step": 58790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5559583834752545e-05,
+      "loss": 1.1955,
+      "step": 58800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555882218532454e-05,
+      "loss": 1.1296,
+      "step": 58810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555806053589654e-05,
+      "loss": 1.1637,
+      "step": 58820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555729888646854e-05,
+      "loss": 1.0915,
+      "step": 58830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5556537237040535e-05,
+      "loss": 1.0448,
+      "step": 58840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555577558761254e-05,
+      "loss": 1.0981,
+      "step": 58850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555501393818453e-05,
+      "loss": 1.2274,
+      "step": 58860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5554252288756534e-05,
+      "loss": 1.2739,
+      "step": 58870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555349063932853e-05,
+      "loss": 1.0946,
+      "step": 58880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555272898990053e-05,
+      "loss": 1.0304,
+      "step": 58890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5551967340472527e-05,
+      "loss": 1.0876,
+      "step": 58900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555120569104453e-05,
+      "loss": 1.0464,
+      "step": 58910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.555044404161652e-05,
+      "loss": 1.0512,
+      "step": 58920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5549682392188526e-05,
+      "loss": 1.0095,
+      "step": 58930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554892074276052e-05,
+      "loss": 1.051,
+      "step": 58940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554815909333252e-05,
+      "loss": 1.1186,
+      "step": 58950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554739744390452e-05,
+      "loss": 0.9363,
+      "step": 58960
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554663579447652e-05,
+      "loss": 1.2065,
+      "step": 58970
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554587414504852e-05,
+      "loss": 1.0608,
+      "step": 58980
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554511249562052e-05,
+      "loss": 1.0624,
+      "step": 58990
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554435084619252e-05,
+      "loss": 1.0994,
+      "step": 59000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.554358919676451e-05,
+      "loss": 1.0138,
+      "step": 59010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5542827547336517e-05,
+      "loss": 1.1849,
+      "step": 59020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5542065897908514e-05,
+      "loss": 1.114,
+      "step": 59030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.554130424848051e-05,
+      "loss": 1.071,
+      "step": 59040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.554054259905251e-05,
+      "loss": 1.1198,
+      "step": 59050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.553978094962451e-05,
+      "loss": 1.0144,
+      "step": 59060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5539019300196504e-05,
+      "loss": 1.141,
+      "step": 59070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.553825765076851e-05,
+      "loss": 1.0112,
+      "step": 59080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5537496001340506e-05,
+      "loss": 1.1627,
+      "step": 59090
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.55367343519125e-05,
+      "loss": 1.1144,
+      "step": 59100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.55359727024845e-05,
+      "loss": 1.1704,
+      "step": 59110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5535211053056505e-05,
+      "loss": 1.0871,
+      "step": 59120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5534449403628495e-05,
+      "loss": 1.2034,
+      "step": 59130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.55336877542005e-05,
+      "loss": 1.0689,
+      "step": 59140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.55329261047725e-05,
+      "loss": 1.1775,
+      "step": 59150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5532164455344495e-05,
+      "loss": 1.2621,
+      "step": 59160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.553140280591649e-05,
+      "loss": 1.1233,
+      "step": 59170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5530641156488496e-05,
+      "loss": 1.0074,
+      "step": 59180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5529879507060494e-05,
+      "loss": 1.0314,
+      "step": 59190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.552911785763249e-05,
+      "loss": 1.0168,
+      "step": 59200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5528356208204495e-05,
+      "loss": 1.0897,
+      "step": 59210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5527594558776486e-05,
+      "loss": 1.1761,
+      "step": 59220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.552683290934849e-05,
+      "loss": 1.1005,
+      "step": 59230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552607125992049e-05,
+      "loss": 1.0811,
+      "step": 59240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5525309610492485e-05,
+      "loss": 1.2061,
+      "step": 59250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552454796106448e-05,
+      "loss": 1.0635,
+      "step": 59260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552378631163649e-05,
+      "loss": 1.0492,
+      "step": 59270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552302466220848e-05,
+      "loss": 1.1336,
+      "step": 59280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552226301278048e-05,
+      "loss": 1.0603,
+      "step": 59290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552150136335247e-05,
+      "loss": 0.9964,
+      "step": 59300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552073971392448e-05,
+      "loss": 0.977,
+      "step": 59310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5519978064496474e-05,
+      "loss": 1.0586,
+      "step": 59320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551921641506847e-05,
+      "loss": 1.1594,
+      "step": 59330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551845476564047e-05,
+      "loss": 1.2133,
+      "step": 59340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5517693116212474e-05,
+      "loss": 1.2112,
+      "step": 59350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551693146678447e-05,
+      "loss": 1.0785,
+      "step": 59360
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551616981735647e-05,
+      "loss": 1.1682,
+      "step": 59370
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551540816792847e-05,
+      "loss": 0.975,
+      "step": 59380
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5514646518500463e-05,
+      "loss": 1.1726,
+      "step": 59390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551388486907247e-05,
+      "loss": 1.1523,
+      "step": 59400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5513123219644465e-05,
+      "loss": 1.1737,
+      "step": 59410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551236157021646e-05,
+      "loss": 1.0354,
+      "step": 59420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.551159992078846e-05,
+      "loss": 1.1807,
+      "step": 59430
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5510838271360464e-05,
+      "loss": 1.1413,
+      "step": 59440
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5510076621932455e-05,
+      "loss": 1.1397,
+      "step": 59450
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550931497250446e-05,
+      "loss": 1.0403,
+      "step": 59460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550855332307646e-05,
+      "loss": 1.031,
+      "step": 59470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5507791673648454e-05,
+      "loss": 1.0637,
+      "step": 59480
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550703002422045e-05,
+      "loss": 0.957,
+      "step": 59490
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5506268374792456e-05,
+      "loss": 1.1288,
+      "step": 59500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5505506725364447e-05,
+      "loss": 1.0879,
+      "step": 59510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550474507593645e-05,
+      "loss": 1.1734,
+      "step": 59520
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550398342650845e-05,
+      "loss": 1.0584,
+      "step": 59530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5503221777080446e-05,
+      "loss": 1.0002,
+      "step": 59540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550246012765244e-05,
+      "loss": 1.0889,
+      "step": 59550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550169847822445e-05,
+      "loss": 0.9747,
+      "step": 59560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5500936828796445e-05,
+      "loss": 1.0066,
+      "step": 59570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.550017517936844e-05,
+      "loss": 1.0616,
+      "step": 59580
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549941352994045e-05,
+      "loss": 1.1151,
+      "step": 59590
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549865188051244e-05,
+      "loss": 1.1519,
+      "step": 59600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549789023108444e-05,
+      "loss": 1.1293,
+      "step": 59610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549712858165644e-05,
+      "loss": 1.0752,
+      "step": 59620
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5496366932228437e-05,
+      "loss": 1.0633,
+      "step": 59630
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5495605282800434e-05,
+      "loss": 1.0894,
+      "step": 59640
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549484363337244e-05,
+      "loss": 1.0523,
+      "step": 59650
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549408198394443e-05,
+      "loss": 1.1516,
+      "step": 59660
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549332033451643e-05,
+      "loss": 1.1962,
+      "step": 59670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.549255868508843e-05,
+      "loss": 1.0813,
+      "step": 59680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.549179703566043e-05,
+      "loss": 1.1221,
+      "step": 59690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5491035386232426e-05,
+      "loss": 1.0904,
+      "step": 59700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.549027373680443e-05,
+      "loss": 1.1872,
+      "step": 59710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548951208737642e-05,
+      "loss": 1.0538,
+      "step": 59720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5488750437948425e-05,
+      "loss": 1.1324,
+      "step": 59730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548798878852042e-05,
+      "loss": 1.0183,
+      "step": 59740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548722713909242e-05,
+      "loss": 1.059,
+      "step": 59750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548646548966442e-05,
+      "loss": 1.0364,
+      "step": 59760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5485703840236415e-05,
+      "loss": 0.9921,
+      "step": 59770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548494219080842e-05,
+      "loss": 1.0622,
+      "step": 59780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5484180541380416e-05,
+      "loss": 0.9389,
+      "step": 59790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5483418891952414e-05,
+      "loss": 1.0893,
+      "step": 59800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548265724252441e-05,
+      "loss": 1.2188,
+      "step": 59810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5481895593096416e-05,
+      "loss": 1.0856,
+      "step": 59820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5481133943668406e-05,
+      "loss": 1.1054,
+      "step": 59830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.548037229424041e-05,
+      "loss": 1.0362,
+      "step": 59840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.547961064481241e-05,
+      "loss": 1.0711,
+      "step": 59850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5478848995384405e-05,
+      "loss": 1.0652,
+      "step": 59860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.54780873459564e-05,
+      "loss": 1.2155,
+      "step": 59870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.547732569652841e-05,
+      "loss": 1.1414,
+      "step": 59880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.54765640471004e-05,
+      "loss": 1.0433,
+      "step": 59890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.54758023976724e-05,
+      "loss": 1.1445,
+      "step": 59900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.54750407482444e-05,
+      "loss": 1.0904,
+      "step": 59910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.54742790988164e-05,
+      "loss": 1.1486,
+      "step": 59920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5473517449388394e-05,
+      "loss": 1.0203,
+      "step": 59930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.54727557999604e-05,
+      "loss": 1.0335,
+      "step": 59940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5471994150532396e-05,
+      "loss": 1.1026,
+      "step": 59950
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5471232501104394e-05,
+      "loss": 1.0121,
+      "step": 59960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.547047085167639e-05,
+      "loss": 1.1543,
+      "step": 59970
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546970920224839e-05,
+      "loss": 1.058,
+      "step": 59980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546894755282039e-05,
+      "loss": 1.0858,
+      "step": 59990
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546818590339239e-05,
+      "loss": 1.11,
+      "step": 60000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546742425396439e-05,
+      "loss": 1.018,
+      "step": 60010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5466662604536385e-05,
+      "loss": 0.9875,
+      "step": 60020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546590095510839e-05,
+      "loss": 1.0504,
+      "step": 60030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546513930568038e-05,
+      "loss": 1.045,
+      "step": 60040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5464377656252384e-05,
+      "loss": 1.2098,
+      "step": 60050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546361600682438e-05,
+      "loss": 1.2141,
+      "step": 60060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546285435739638e-05,
+      "loss": 0.9163,
+      "step": 60070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546209270796838e-05,
+      "loss": 1.27,
+      "step": 60080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546133105854038e-05,
+      "loss": 0.9151,
+      "step": 60090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.546056940911237e-05,
+      "loss": 1.0457,
+      "step": 60100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5459807759684376e-05,
+      "loss": 1.023,
+      "step": 60110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5459046110256373e-05,
+      "loss": 1.1336,
+      "step": 60120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545828446082837e-05,
+      "loss": 1.1013,
+      "step": 60130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545752281140037e-05,
+      "loss": 1.0337,
+      "step": 60140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545676116197237e-05,
+      "loss": 1.0917,
+      "step": 60150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545599951254437e-05,
+      "loss": 1.0718,
+      "step": 60160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545523786311637e-05,
+      "loss": 1.18,
+      "step": 60170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545447621368837e-05,
+      "loss": 0.9517,
+      "step": 60180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545371456426036e-05,
+      "loss": 1.0124,
+      "step": 60190
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545295291483237e-05,
+      "loss": 1.1178,
+      "step": 60200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545219126540436e-05,
+      "loss": 1.1175,
+      "step": 60210
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545142961597636e-05,
+      "loss": 1.0675,
+      "step": 60220
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.545066796654836e-05,
+      "loss": 1.1326,
+      "step": 60230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5449906317120357e-05,
+      "loss": 1.093,
+      "step": 60240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5449144667692354e-05,
+      "loss": 1.0486,
+      "step": 60250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544838301826436e-05,
+      "loss": 1.0681,
+      "step": 60260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544762136883635e-05,
+      "loss": 1.1144,
+      "step": 60270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544685971940835e-05,
+      "loss": 1.114,
+      "step": 60280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544609806998035e-05,
+      "loss": 1.081,
+      "step": 60290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544533642055235e-05,
+      "loss": 1.1092,
+      "step": 60300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5444574771124346e-05,
+      "loss": 0.992,
+      "step": 60310
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544381312169635e-05,
+      "loss": 1.0899,
+      "step": 60320
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.544305147226835e-05,
+      "loss": 1.0083,
+      "step": 60330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5442289822840345e-05,
+      "loss": 1.1409,
+      "step": 60340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.544152817341234e-05,
+      "loss": 1.1781,
+      "step": 60350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.544076652398434e-05,
+      "loss": 1.0155,
+      "step": 60360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5440004874556344e-05,
+      "loss": 1.0142,
+      "step": 60370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543924322512834e-05,
+      "loss": 1.2564,
+      "step": 60380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543848157570034e-05,
+      "loss": 1.0702,
+      "step": 60390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5437719926272336e-05,
+      "loss": 1.0586,
+      "step": 60400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543695827684434e-05,
+      "loss": 1.0046,
+      "step": 60410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543619662741633e-05,
+      "loss": 1.0659,
+      "step": 60420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5435434977988336e-05,
+      "loss": 1.2206,
+      "step": 60430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543467332856033e-05,
+      "loss": 1.1776,
+      "step": 60440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543391167913233e-05,
+      "loss": 1.0897,
+      "step": 60450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543315002970433e-05,
+      "loss": 1.1007,
+      "step": 60460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543238838027633e-05,
+      "loss": 1.0927,
+      "step": 60470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543162673084832e-05,
+      "loss": 1.0096,
+      "step": 60480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.543086508142033e-05,
+      "loss": 1.0268,
+      "step": 60490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5430103431992325e-05,
+      "loss": 1.1709,
+      "step": 60500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.542934178256432e-05,
+      "loss": 1.1478,
+      "step": 60510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.542858013313632e-05,
+      "loss": 1.0321,
+      "step": 60520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5427818483708324e-05,
+      "loss": 1.1662,
+      "step": 60530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.542705683428032e-05,
+      "loss": 1.078,
+      "step": 60540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.542629518485232e-05,
+      "loss": 1.1581,
+      "step": 60550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5425533535424316e-05,
+      "loss": 1.1589,
+      "step": 60560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5424771885996314e-05,
+      "loss": 1.1282,
+      "step": 60570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.542401023656832e-05,
+      "loss": 0.9838,
+      "step": 60580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5423248587140315e-05,
+      "loss": 1.1894,
+      "step": 60590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.542248693771231e-05,
+      "loss": 0.9567,
+      "step": 60600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.542172528828431e-05,
+      "loss": 1.0421,
+      "step": 60610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5420963638856314e-05,
+      "loss": 1.1374,
+      "step": 60620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5420201989428305e-05,
+      "loss": 1.1501,
+      "step": 60630
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.541944034000031e-05,
+      "loss": 1.1607,
+      "step": 60640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.541867869057231e-05,
+      "loss": 1.1819,
+      "step": 60650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5417917041144304e-05,
+      "loss": 1.1461,
+      "step": 60660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.54171553917163e-05,
+      "loss": 0.9883,
+      "step": 60670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.54163937422883e-05,
+      "loss": 1.0833,
+      "step": 60680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.54156320928603e-05,
+      "loss": 1.0374,
+      "step": 60690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.54148704434323e-05,
+      "loss": 0.9839,
+      "step": 60700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.54141087940043e-05,
+      "loss": 1.2199,
+      "step": 60710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5413347144576296e-05,
+      "loss": 1.1864,
+      "step": 60720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5412585495148293e-05,
+      "loss": 1.0685,
+      "step": 60730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.541182384572029e-05,
+      "loss": 1.0821,
+      "step": 60740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5411062196292295e-05,
+      "loss": 1.2154,
+      "step": 60750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.541030054686429e-05,
+      "loss": 1.1532,
+      "step": 60760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.540953889743629e-05,
+      "loss": 1.103,
+      "step": 60770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540877724800829e-05,
+      "loss": 1.0193,
+      "step": 60780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540801559858029e-05,
+      "loss": 1.2312,
+      "step": 60790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540725394915228e-05,
+      "loss": 1.1022,
+      "step": 60800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540649229972429e-05,
+      "loss": 1.0109,
+      "step": 60810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5405730650296284e-05,
+      "loss": 1.0977,
+      "step": 60820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540496900086828e-05,
+      "loss": 0.9806,
+      "step": 60830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540420735144028e-05,
+      "loss": 1.0998,
+      "step": 60840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540344570201228e-05,
+      "loss": 1.076,
+      "step": 60850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5402684052584274e-05,
+      "loss": 1.064,
+      "step": 60860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540192240315628e-05,
+      "loss": 1.1471,
+      "step": 60870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5401160753728276e-05,
+      "loss": 1.1209,
+      "step": 60880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.540039910430027e-05,
+      "loss": 1.2216,
+      "step": 60890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.539963745487227e-05,
+      "loss": 1.091,
+      "step": 60900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5398875805444275e-05,
+      "loss": 1.0911,
+      "step": 60910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.539811415601627e-05,
+      "loss": 1.0886,
+      "step": 60920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.539735250658827e-05,
+      "loss": 1.1541,
+      "step": 60930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.539659085716027e-05,
+      "loss": 1.2374,
+      "step": 60940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5395829207732265e-05,
+      "loss": 1.1902,
+      "step": 60950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.539506755830427e-05,
+      "loss": 1.0117,
+      "step": 60960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5394305908876266e-05,
+      "loss": 0.9849,
+      "step": 60970
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5393544259448264e-05,
+      "loss": 1.0369,
+      "step": 60980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.539278261002026e-05,
+      "loss": 1.0459,
+      "step": 60990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5392020960592266e-05,
+      "loss": 1.029,
+      "step": 61000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5391259311164256e-05,
+      "loss": 1.0922,
+      "step": 61010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.539049766173626e-05,
+      "loss": 1.202,
+      "step": 61020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538973601230826e-05,
+      "loss": 0.8779,
+      "step": 61030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5388974362880256e-05,
+      "loss": 1.1579,
+      "step": 61040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538821271345225e-05,
+      "loss": 1.1798,
+      "step": 61050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538745106402426e-05,
+      "loss": 1.0344,
+      "step": 61060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538668941459625e-05,
+      "loss": 1.2077,
+      "step": 61070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538592776516825e-05,
+      "loss": 1.1718,
+      "step": 61080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538516611574025e-05,
+      "loss": 1.0054,
+      "step": 61090
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538440446631225e-05,
+      "loss": 1.0766,
+      "step": 61100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5383642816884245e-05,
+      "loss": 1.0657,
+      "step": 61110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538288116745625e-05,
+      "loss": 1.0909,
+      "step": 61120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5382119518028246e-05,
+      "loss": 1.0719,
+      "step": 61130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5381357868600244e-05,
+      "loss": 1.1112,
+      "step": 61140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.538059621917224e-05,
+      "loss": 1.0421,
+      "step": 61150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.537983456974424e-05,
+      "loss": 1.0503,
+      "step": 61160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.537907292031624e-05,
+      "loss": 1.1821,
+      "step": 61170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5378311270888234e-05,
+      "loss": 0.9932,
+      "step": 61180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.537754962146024e-05,
+      "loss": 1.0294,
+      "step": 61190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5376787972032235e-05,
+      "loss": 1.0731,
+      "step": 61200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.537602632260423e-05,
+      "loss": 1.1712,
+      "step": 61210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.537526467317623e-05,
+      "loss": 1.0145,
+      "step": 61220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5374503023748235e-05,
+      "loss": 1.1096,
+      "step": 61230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5373741374320225e-05,
+      "loss": 1.1284,
+      "step": 61240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.537297972489223e-05,
+      "loss": 1.1251,
+      "step": 61250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.537221807546423e-05,
+      "loss": 1.1011,
+      "step": 61260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5371456426036224e-05,
+      "loss": 1.0121,
+      "step": 61270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.537069477660822e-05,
+      "loss": 1.0298,
+      "step": 61280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5369933127180226e-05,
+      "loss": 1.0811,
+      "step": 61290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5369171477752224e-05,
+      "loss": 1.0438,
+      "step": 61300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536840982832422e-05,
+      "loss": 1.0317,
+      "step": 61310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536764817889622e-05,
+      "loss": 0.969,
+      "step": 61320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5366886529468216e-05,
+      "loss": 1.1219,
+      "step": 61330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536612488004022e-05,
+      "loss": 1.1208,
+      "step": 61340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536536323061222e-05,
+      "loss": 1.1991,
+      "step": 61350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5364601581184215e-05,
+      "loss": 1.0959,
+      "step": 61360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536383993175621e-05,
+      "loss": 1.1994,
+      "step": 61370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536307828232822e-05,
+      "loss": 1.1624,
+      "step": 61380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536231663290021e-05,
+      "loss": 1.041,
+      "step": 61390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536155498347221e-05,
+      "loss": 1.1187,
+      "step": 61400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536079333404421e-05,
+      "loss": 1.049,
+      "step": 61410
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.536003168461621e-05,
+      "loss": 1.1537,
+      "step": 61420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5359270035188204e-05,
+      "loss": 1.1382,
+      "step": 61430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.535850838576021e-05,
+      "loss": 1.0154,
+      "step": 61440
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.53577467363322e-05,
+      "loss": 1.0839,
+      "step": 61450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.53569850869042e-05,
+      "loss": 0.986,
+      "step": 61460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.53562234374762e-05,
+      "loss": 1.2279,
+      "step": 61470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.53554617880482e-05,
+      "loss": 1.0665,
+      "step": 61480
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5354700138620196e-05,
+      "loss": 1.1577,
+      "step": 61490
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.53539384891922e-05,
+      "loss": 1.0405,
+      "step": 61500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.53531768397642e-05,
+      "loss": 1.0616,
+      "step": 61510
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5352415190336195e-05,
+      "loss": 1.101,
+      "step": 61520
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.535165354090819e-05,
+      "loss": 1.352,
+      "step": 61530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.535089189148019e-05,
+      "loss": 1.1557,
+      "step": 61540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5350130242052194e-05,
+      "loss": 1.0599,
+      "step": 61550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.534936859262419e-05,
+      "loss": 1.0604,
+      "step": 61560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.534860694319619e-05,
+      "loss": 1.0905,
+      "step": 61570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5347845293768186e-05,
+      "loss": 1.0704,
+      "step": 61580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5347083644340184e-05,
+      "loss": 1.1156,
+      "step": 61590
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.534632199491218e-05,
+      "loss": 1.0777,
+      "step": 61600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5345560345484186e-05,
+      "loss": 1.0602,
+      "step": 61610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5344798696056176e-05,
+      "loss": 0.9498,
+      "step": 61620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.534403704662818e-05,
+      "loss": 1.1339,
+      "step": 61630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.534327539720018e-05,
+      "loss": 1.1049,
+      "step": 61640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5342513747772176e-05,
+      "loss": 1.1124,
+      "step": 61650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.534175209834417e-05,
+      "loss": 1.0195,
+      "step": 61660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.534099044891618e-05,
+      "loss": 1.054,
+      "step": 61670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5340228799488175e-05,
+      "loss": 1.0814,
+      "step": 61680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533946715006017e-05,
+      "loss": 1.137,
+      "step": 61690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533870550063217e-05,
+      "loss": 0.9849,
+      "step": 61700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533794385120417e-05,
+      "loss": 1.1241,
+      "step": 61710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533718220177617e-05,
+      "loss": 1.0006,
+      "step": 61720
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533642055234817e-05,
+      "loss": 1.2117,
+      "step": 61730
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5335658902920166e-05,
+      "loss": 1.0123,
+      "step": 61740
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5334897253492164e-05,
+      "loss": 1.0623,
+      "step": 61750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533413560406417e-05,
+      "loss": 1.031,
+      "step": 61760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533337395463616e-05,
+      "loss": 1.0192,
+      "step": 61770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533261230520816e-05,
+      "loss": 1.1112,
+      "step": 61780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533185065578016e-05,
+      "loss": 1.1129,
+      "step": 61790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.533108900635216e-05,
+      "loss": 1.0519,
+      "step": 61800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5330327356924155e-05,
+      "loss": 1.1148,
+      "step": 61810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.532956570749616e-05,
+      "loss": 0.9467,
+      "step": 61820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.532880405806815e-05,
+      "loss": 1.0946,
+      "step": 61830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5328042408640155e-05,
+      "loss": 1.0961,
+      "step": 61840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.532728075921215e-05,
+      "loss": 1.1364,
+      "step": 61850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.532651910978415e-05,
+      "loss": 0.9883,
+      "step": 61860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.532575746035615e-05,
+      "loss": 1.0248,
+      "step": 61870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.532499581092815e-05,
+      "loss": 0.9947,
+      "step": 61880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.532423416150015e-05,
+      "loss": 1.1409,
+      "step": 61890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.5323472512072146e-05,
+      "loss": 1.1655,
+      "step": 61900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.5322710862644144e-05,
+      "loss": 1.0545,
+      "step": 61910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.532194921321614e-05,
+      "loss": 1.0898,
+      "step": 61920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.5321187563788145e-05,
+      "loss": 1.0975,
+      "step": 61930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.532042591436014e-05,
+      "loss": 1.0181,
+      "step": 61940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531966426493214e-05,
+      "loss": 0.9543,
+      "step": 61950
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531890261550414e-05,
+      "loss": 1.0219,
+      "step": 61960
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531814096607614e-05,
+      "loss": 1.01,
+      "step": 61970
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531737931664813e-05,
+      "loss": 1.1819,
+      "step": 61980
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531661766722014e-05,
+      "loss": 1.1425,
+      "step": 61990
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.5315856017792134e-05,
+      "loss": 1.1394,
+      "step": 62000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531509436836413e-05,
+      "loss": 1.0657,
+      "step": 62010
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531433271893613e-05,
+      "loss": 1.113,
+      "step": 62020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.5313571069508133e-05,
+      "loss": 1.1413,
+      "step": 62030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.5312809420080124e-05,
+      "loss": 0.9937,
+      "step": 62040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531204777065213e-05,
+      "loss": 1.0181,
+      "step": 62050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531128612122412e-05,
+      "loss": 1.2271,
+      "step": 62060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.531052447179612e-05,
+      "loss": 1.0565,
+      "step": 62070
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.530976282236812e-05,
+      "loss": 1.0568,
+      "step": 62080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530900117294012e-05,
+      "loss": 1.0698,
+      "step": 62090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530823952351212e-05,
+      "loss": 0.9886,
+      "step": 62100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530747787408412e-05,
+      "loss": 1.0948,
+      "step": 62110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530671622465612e-05,
+      "loss": 1.1176,
+      "step": 62120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.5305954575228115e-05,
+      "loss": 0.9506,
+      "step": 62130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530519292580012e-05,
+      "loss": 1.0601,
+      "step": 62140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530443127637211e-05,
+      "loss": 1.1483,
+      "step": 62150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.5303669626944114e-05,
+      "loss": 1.0703,
+      "step": 62160
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530290797751611e-05,
+      "loss": 1.1339,
+      "step": 62170
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530214632808811e-05,
+      "loss": 0.9911,
+      "step": 62180
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.5301384678660107e-05,
+      "loss": 1.1081,
+      "step": 62190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.530062302923211e-05,
+      "loss": 1.1287,
+      "step": 62200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.52998613798041e-05,
+      "loss": 1.0598,
+      "step": 62210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.5299099730376106e-05,
+      "loss": 1.2035,
+      "step": 62220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.52983380809481e-05,
+      "loss": 1.161,
+      "step": 62230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.52975764315201e-05,
+      "loss": 1.1162,
+      "step": 62240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.52968147820921e-05,
+      "loss": 1.115,
+      "step": 62250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.52960531326641e-05,
+      "loss": 1.11,
+      "step": 62260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.529529148323609e-05,
+      "loss": 1.0843,
+      "step": 62270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.52945298338081e-05,
+      "loss": 1.0418,
+      "step": 62280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.5293768184380095e-05,
+      "loss": 1.1502,
+      "step": 62290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.529300653495209e-05,
+      "loss": 1.0408,
+      "step": 62300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5292244885524096e-05,
+      "loss": 1.0789,
+      "step": 62310
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5291483236096094e-05,
+      "loss": 1.1599,
+      "step": 62320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.529072158666809e-05,
+      "loss": 1.1297,
+      "step": 62330
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528995993724009e-05,
+      "loss": 1.0545,
+      "step": 62340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528919828781209e-05,
+      "loss": 1.1571,
+      "step": 62350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5288436638384084e-05,
+      "loss": 1.1466,
+      "step": 62360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528767498895609e-05,
+      "loss": 1.0323,
+      "step": 62370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5286913339528085e-05,
+      "loss": 1.0681,
+      "step": 62380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528615169010008e-05,
+      "loss": 1.1006,
+      "step": 62390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528539004067208e-05,
+      "loss": 1.1619,
+      "step": 62400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5284628391244085e-05,
+      "loss": 1.0713,
+      "step": 62410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5283866741816075e-05,
+      "loss": 1.0574,
+      "step": 62420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528310509238808e-05,
+      "loss": 1.0639,
+      "step": 62430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528234344296008e-05,
+      "loss": 1.1051,
+      "step": 62440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5281581793532075e-05,
+      "loss": 1.0545,
+      "step": 62450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.528082014410407e-05,
+      "loss": 1.0582,
+      "step": 62460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5280058494676076e-05,
+      "loss": 1.0563,
+      "step": 62470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5279296845248074e-05,
+      "loss": 0.9447,
+      "step": 62480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.527853519582007e-05,
+      "loss": 0.9997,
+      "step": 62490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.527777354639207e-05,
+      "loss": 1.1629,
+      "step": 62500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.5277011896964066e-05,
+      "loss": 0.9569,
+      "step": 62510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.527625024753607e-05,
+      "loss": 1.2235,
+      "step": 62520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527548859810806e-05,
+      "loss": 1.1233,
+      "step": 62530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5274726948680065e-05,
+      "loss": 1.1147,
+      "step": 62540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527396529925206e-05,
+      "loss": 1.0234,
+      "step": 62550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527320364982406e-05,
+      "loss": 1.1157,
+      "step": 62560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527244200039606e-05,
+      "loss": 1.0859,
+      "step": 62570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527168035096806e-05,
+      "loss": 1.0877,
+      "step": 62580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527091870154005e-05,
+      "loss": 1.0775,
+      "step": 62590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.527015705211206e-05,
+      "loss": 1.1738,
+      "step": 62600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5269395402684054e-05,
+      "loss": 1.0496,
+      "step": 62610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526863375325605e-05,
+      "loss": 1.055,
+      "step": 62620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526787210382805e-05,
+      "loss": 0.9541,
+      "step": 62630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5267110454400053e-05,
+      "loss": 1.0179,
+      "step": 62640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5266348804972044e-05,
+      "loss": 1.0601,
+      "step": 62650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526558715554405e-05,
+      "loss": 1.042,
+      "step": 62660
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5264825506116046e-05,
+      "loss": 1.102,
+      "step": 62670
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526406385668804e-05,
+      "loss": 1.0652,
+      "step": 62680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526330220726005e-05,
+      "loss": 1.0942,
+      "step": 62690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5262540557832045e-05,
+      "loss": 1.1315,
+      "step": 62700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526177890840404e-05,
+      "loss": 1.0393,
+      "step": 62710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.526101725897604e-05,
+      "loss": 1.1371,
+      "step": 62720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5260255609548044e-05,
+      "loss": 1.109,
+      "step": 62730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5259493960120035e-05,
+      "loss": 1.0228,
+      "step": 62740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525873231069204e-05,
+      "loss": 1.221,
+      "step": 62750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525797066126404e-05,
+      "loss": 1.0849,
+      "step": 62760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5257209011836034e-05,
+      "loss": 1.0699,
+      "step": 62770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525644736240803e-05,
+      "loss": 1.0572,
+      "step": 62780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5255685712980036e-05,
+      "loss": 1.0729,
+      "step": 62790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5254924063552027e-05,
+      "loss": 1.0824,
+      "step": 62800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525416241412403e-05,
+      "loss": 1.0258,
+      "step": 62810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525340076469603e-05,
+      "loss": 1.1691,
+      "step": 62820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5252639115268026e-05,
+      "loss": 0.9548,
+      "step": 62830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525187746584002e-05,
+      "loss": 1.1293,
+      "step": 62840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525111581641203e-05,
+      "loss": 1.0742,
+      "step": 62850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.525035416698402e-05,
+      "loss": 1.1042,
+      "step": 62860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524959251755602e-05,
+      "loss": 1.187,
+      "step": 62870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524883086812802e-05,
+      "loss": 1.0741,
+      "step": 62880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524806921870002e-05,
+      "loss": 1.166,
+      "step": 62890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524730756927202e-05,
+      "loss": 1.1287,
+      "step": 62900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524654591984402e-05,
+      "loss": 1.1267,
+      "step": 62910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5245784270416016e-05,
+      "loss": 0.9754,
+      "step": 62920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.5245022620988014e-05,
+      "loss": 1.0867,
+      "step": 62930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524426097156002e-05,
+      "loss": 1.0185,
+      "step": 62940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524349932213201e-05,
+      "loss": 1.2207,
+      "step": 62950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.524273767270401e-05,
+      "loss": 1.3101,
+      "step": 62960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5241976023276004e-05,
+      "loss": 1.0754,
+      "step": 62970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.524121437384801e-05,
+      "loss": 1.2578,
+      "step": 62980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5240452724420005e-05,
+      "loss": 1.0877,
+      "step": 62990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5239691074992e-05,
+      "loss": 1.0949,
+      "step": 63000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5238929425564e-05,
+      "loss": 1.1221,
+      "step": 63010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5238167776136005e-05,
+      "loss": 1.0601,
+      "step": 63020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5237406126707995e-05,
+      "loss": 0.9626,
+      "step": 63030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.523664447728e-05,
+      "loss": 1.013,
+      "step": 63040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5235882827852e-05,
+      "loss": 0.9747,
+      "step": 63050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5235121178423995e-05,
+      "loss": 1.0125,
+      "step": 63060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.523435952899599e-05,
+      "loss": 1.0808,
+      "step": 63070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5233597879567996e-05,
+      "loss": 1.0802,
+      "step": 63080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5232836230139994e-05,
+      "loss": 0.9699,
+      "step": 63090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.523207458071199e-05,
+      "loss": 0.9946,
+      "step": 63100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5231312931283995e-05,
+      "loss": 1.0002,
+      "step": 63110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5230551281855986e-05,
+      "loss": 1.1353,
+      "step": 63120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.522978963242799e-05,
+      "loss": 1.0162,
+      "step": 63130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.522902798299999e-05,
+      "loss": 0.9039,
+      "step": 63140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5228266333571985e-05,
+      "loss": 1.156,
+      "step": 63150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.522750468414398e-05,
+      "loss": 1.0754,
+      "step": 63160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.522674303471599e-05,
+      "loss": 1.0076,
+      "step": 63170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.522598138528798e-05,
+      "loss": 0.9489,
+      "step": 63180
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.522521973585998e-05,
+      "loss": 1.088,
+      "step": 63190
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.522445808643198e-05,
+      "loss": 1.0596,
+      "step": 63200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.522369643700398e-05,
+      "loss": 1.1009,
+      "step": 63210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5222934787575974e-05,
+      "loss": 1.0665,
+      "step": 63220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.522217313814798e-05,
+      "loss": 0.9382,
+      "step": 63230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.522141148871997e-05,
+      "loss": 1.0853,
+      "step": 63240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5220649839291974e-05,
+      "loss": 0.9952,
+      "step": 63250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521988818986397e-05,
+      "loss": 1.0675,
+      "step": 63260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521912654043597e-05,
+      "loss": 1.0841,
+      "step": 63270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521836489100797e-05,
+      "loss": 1.1726,
+      "step": 63280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521760324157997e-05,
+      "loss": 1.1693,
+      "step": 63290
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521684159215197e-05,
+      "loss": 1.1509,
+      "step": 63300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5216079942723965e-05,
+      "loss": 1.0515,
+      "step": 63310
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521531829329597e-05,
+      "loss": 1.0042,
+      "step": 63320
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521455664386796e-05,
+      "loss": 1.03,
+      "step": 63330
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.5213794994439964e-05,
+      "loss": 0.9859,
+      "step": 63340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521303334501196e-05,
+      "loss": 1.1497,
+      "step": 63350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521227169558396e-05,
+      "loss": 1.0854,
+      "step": 63360
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521151004615596e-05,
+      "loss": 1.1733,
+      "step": 63370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.521074839672796e-05,
+      "loss": 1.0849,
+      "step": 63380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.520998674729995e-05,
+      "loss": 1.0942,
+      "step": 63390
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5209225097871956e-05,
+      "loss": 1.1894,
+      "step": 63400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520846344844395e-05,
+      "loss": 1.02,
+      "step": 63410
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520770179901595e-05,
+      "loss": 1.1065,
+      "step": 63420
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520694014958795e-05,
+      "loss": 0.9597,
+      "step": 63430
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5206178500159946e-05,
+      "loss": 1.0619,
+      "step": 63440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520541685073194e-05,
+      "loss": 1.1559,
+      "step": 63450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520465520130395e-05,
+      "loss": 1.1321,
+      "step": 63460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5203893551875945e-05,
+      "loss": 1.1843,
+      "step": 63470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520313190244794e-05,
+      "loss": 1.2482,
+      "step": 63480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5202370253019947e-05,
+      "loss": 1.0497,
+      "step": 63490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520160860359194e-05,
+      "loss": 1.0333,
+      "step": 63500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520084695416394e-05,
+      "loss": 1.1546,
+      "step": 63510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.520008530473594e-05,
+      "loss": 1.0903,
+      "step": 63520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5199323655307936e-05,
+      "loss": 0.9366,
+      "step": 63530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5198562005879934e-05,
+      "loss": 1.0029,
+      "step": 63540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.519780035645194e-05,
+      "loss": 1.0317,
+      "step": 63550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.519703870702393e-05,
+      "loss": 1.0623,
+      "step": 63560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.519627705759593e-05,
+      "loss": 1.1565,
+      "step": 63570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.519551540816793e-05,
+      "loss": 1.0232,
+      "step": 63580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.519475375873993e-05,
+      "loss": 1.3425,
+      "step": 63590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.5193992109311926e-05,
+      "loss": 1.1472,
+      "step": 63600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.519323045988393e-05,
+      "loss": 1.0591,
+      "step": 63610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.519246881045592e-05,
+      "loss": 1.088,
+      "step": 63620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5191707161027925e-05,
+      "loss": 1.0874,
+      "step": 63630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.519094551159992e-05,
+      "loss": 1.0591,
+      "step": 63640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.519018386217192e-05,
+      "loss": 1.0751,
+      "step": 63650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518942221274392e-05,
+      "loss": 0.9179,
+      "step": 63660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518866056331592e-05,
+      "loss": 1.0726,
+      "step": 63670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518789891388792e-05,
+      "loss": 1.0878,
+      "step": 63680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5187137264459916e-05,
+      "loss": 1.054,
+      "step": 63690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518637561503192e-05,
+      "loss": 1.0178,
+      "step": 63700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518561396560391e-05,
+      "loss": 1.2013,
+      "step": 63710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5184852316175915e-05,
+      "loss": 1.1024,
+      "step": 63720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518409066674791e-05,
+      "loss": 1.075,
+      "step": 63730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518332901731991e-05,
+      "loss": 1.069,
+      "step": 63740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518256736789191e-05,
+      "loss": 1.1516,
+      "step": 63750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518180571846391e-05,
+      "loss": 1.011,
+      "step": 63760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.51810440690359e-05,
+      "loss": 1.1378,
+      "step": 63770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.518028241960791e-05,
+      "loss": 1.1132,
+      "step": 63780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5179520770179904e-05,
+      "loss": 1.1035,
+      "step": 63790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.51787591207519e-05,
+      "loss": 1.0358,
+      "step": 63800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.51779974713239e-05,
+      "loss": 1.3327,
+      "step": 63810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5177235821895904e-05,
+      "loss": 1.0419,
+      "step": 63820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5176474172467894e-05,
+      "loss": 1.1038,
+      "step": 63830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.51757125230399e-05,
+      "loss": 1.1484,
+      "step": 63840
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5174950873611896e-05,
+      "loss": 0.9361,
+      "step": 63850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5174189224183894e-05,
+      "loss": 1.0536,
+      "step": 63860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.51734275747559e-05,
+      "loss": 1.0583,
+      "step": 63870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5172665925327895e-05,
+      "loss": 1.1002,
+      "step": 63880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.517190427589989e-05,
+      "loss": 0.9945,
+      "step": 63890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.517114262647189e-05,
+      "loss": 1.0457,
+      "step": 63900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.517038097704389e-05,
+      "loss": 1.0555,
+      "step": 63910
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5169619327615885e-05,
+      "loss": 1.0249,
+      "step": 63920
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516885767818789e-05,
+      "loss": 1.2193,
+      "step": 63930
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516809602875988e-05,
+      "loss": 1.0597,
+      "step": 63940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5167334379331884e-05,
+      "loss": 1.127,
+      "step": 63950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516657272990388e-05,
+      "loss": 1.0488,
+      "step": 63960
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516581108047588e-05,
+      "loss": 1.1586,
+      "step": 63970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516504943104788e-05,
+      "loss": 1.0351,
+      "step": 63980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516428778161988e-05,
+      "loss": 1.0089,
+      "step": 63990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516352613219187e-05,
+      "loss": 1.0247,
+      "step": 64000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5162764482763876e-05,
+      "loss": 1.117,
+      "step": 64010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516200283333587e-05,
+      "loss": 1.0391,
+      "step": 64020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516124118390787e-05,
+      "loss": 1.0589,
+      "step": 64030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.516047953447987e-05,
+      "loss": 1.1269,
+      "step": 64040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.515971788505187e-05,
+      "loss": 1.0969,
+      "step": 64050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515895623562387e-05,
+      "loss": 1.1108,
+      "step": 64060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515819458619587e-05,
+      "loss": 1.1475,
+      "step": 64070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515743293676787e-05,
+      "loss": 1.0842,
+      "step": 64080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515667128733986e-05,
+      "loss": 1.1402,
+      "step": 64090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5155909637911867e-05,
+      "loss": 1.0906,
+      "step": 64100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5155147988483864e-05,
+      "loss": 1.0257,
+      "step": 64110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515438633905586e-05,
+      "loss": 1.1317,
+      "step": 64120
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515362468962786e-05,
+      "loss": 1.0161,
+      "step": 64130
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515286304019986e-05,
+      "loss": 0.9935,
+      "step": 64140
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5152101390771854e-05,
+      "loss": 1.0642,
+      "step": 64150
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.515133974134386e-05,
+      "loss": 1.2272,
+      "step": 64160
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5150578091915856e-05,
+      "loss": 1.0763,
+      "step": 64170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.514981644248785e-05,
+      "loss": 1.0643,
+      "step": 64180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.514905479305985e-05,
+      "loss": 1.132,
+      "step": 64190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5148293143631855e-05,
+      "loss": 1.2065,
+      "step": 64200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5147531494203846e-05,
+      "loss": 1.0834,
+      "step": 64210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.514676984477585e-05,
+      "loss": 1.191,
+      "step": 64220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.514600819534785e-05,
+      "loss": 1.1367,
+      "step": 64230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5145246545919845e-05,
+      "loss": 1.1636,
+      "step": 64240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.514448489649184e-05,
+      "loss": 1.0906,
+      "step": 64250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5143723247063846e-05,
+      "loss": 1.1023,
+      "step": 64260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5142961597635844e-05,
+      "loss": 1.1207,
+      "step": 64270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.514219994820784e-05,
+      "loss": 1.0613,
+      "step": 64280
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5141438298779846e-05,
+      "loss": 1.0679,
+      "step": 64290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5140676649351836e-05,
+      "loss": 1.0997,
+      "step": 64300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513991499992384e-05,
+      "loss": 1.0586,
+      "step": 64310
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513915335049584e-05,
+      "loss": 0.9914,
+      "step": 64320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5138391701067835e-05,
+      "loss": 1.1135,
+      "step": 64330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513763005163983e-05,
+      "loss": 1.2188,
+      "step": 64340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513686840221183e-05,
+      "loss": 1.003,
+      "step": 64350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513610675278383e-05,
+      "loss": 1.1409,
+      "step": 64360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513534510335583e-05,
+      "loss": 1.0773,
+      "step": 64370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513458345392782e-05,
+      "loss": 1.1495,
+      "step": 64380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513382180449983e-05,
+      "loss": 1.0977,
+      "step": 64390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5133060155071824e-05,
+      "loss": 1.0128,
+      "step": 64400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513229850564382e-05,
+      "loss": 1.0269,
+      "step": 64410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513153685621582e-05,
+      "loss": 1.1105,
+      "step": 64420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5130775206787824e-05,
+      "loss": 1.1223,
+      "step": 64430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.513001355735982e-05,
+      "loss": 1.1597,
+      "step": 64440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.512925190793182e-05,
+      "loss": 1.222,
+      "step": 64450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5128490258503816e-05,
+      "loss": 0.9667,
+      "step": 64460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5127728609075814e-05,
+      "loss": 1.1682,
+      "step": 64470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.512696695964782e-05,
+      "loss": 1.0766,
+      "step": 64480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5126205310219815e-05,
+      "loss": 1.0695,
+      "step": 64490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.512544366079181e-05,
+      "loss": 1.0434,
+      "step": 64500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.512468201136381e-05,
+      "loss": 0.9648,
+      "step": 64510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5123920361935814e-05,
+      "loss": 1.0744,
+      "step": 64520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5123158712507805e-05,
+      "loss": 1.1452,
+      "step": 64530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.512239706307981e-05,
+      "loss": 1.1426,
+      "step": 64540
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.512163541365181e-05,
+      "loss": 1.23,
+      "step": 64550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5120873764223804e-05,
+      "loss": 1.142,
+      "step": 64560
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.51201121147958e-05,
+      "loss": 1.1137,
+      "step": 64570
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5119350465367806e-05,
+      "loss": 1.0625,
+      "step": 64580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.51185888159398e-05,
+      "loss": 1.1139,
+      "step": 64590
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.51178271665118e-05,
+      "loss": 1.0638,
+      "step": 64600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.51170655170838e-05,
+      "loss": 0.9022,
+      "step": 64610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5116303867655796e-05,
+      "loss": 1.0476,
+      "step": 64620
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.511554221822779e-05,
+      "loss": 0.9884,
+      "step": 64630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.51147805687998e-05,
+      "loss": 1.0874,
+      "step": 64640
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.5114018919371795e-05,
+      "loss": 1.1492,
+      "step": 64650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.511325726994379e-05,
+      "loss": 1.0311,
+      "step": 64660
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.51124956205158e-05,
+      "loss": 0.9792,
+      "step": 64670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.511173397108779e-05,
+      "loss": 1.1093,
+      "step": 64680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.511097232165979e-05,
+      "loss": 1.0711,
+      "step": 64690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.511021067223179e-05,
+      "loss": 1.1907,
+      "step": 64700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.510944902280379e-05,
+      "loss": 1.0129,
+      "step": 64710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5108687373375784e-05,
+      "loss": 1.0212,
+      "step": 64720
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510792572394779e-05,
+      "loss": 1.1452,
+      "step": 64730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510716407451978e-05,
+      "loss": 1.1729,
+      "step": 64740
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510640242509178e-05,
+      "loss": 0.9732,
+      "step": 64750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510564077566378e-05,
+      "loss": 0.9911,
+      "step": 64760
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510487912623578e-05,
+      "loss": 1.1227,
+      "step": 64770
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5104117476807776e-05,
+      "loss": 1.0544,
+      "step": 64780
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510335582737978e-05,
+      "loss": 1.0763,
+      "step": 64790
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510259417795177e-05,
+      "loss": 1.0863,
+      "step": 64800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5101832528523775e-05,
+      "loss": 1.1408,
+      "step": 64810
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510107087909577e-05,
+      "loss": 1.0159,
+      "step": 64820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.510030922966777e-05,
+      "loss": 1.1448,
+      "step": 64830
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.509954758023977e-05,
+      "loss": 0.9228,
+      "step": 64840
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5098785930811765e-05,
+      "loss": 1.0947,
+      "step": 64850
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.509802428138377e-05,
+      "loss": 1.2274,
+      "step": 64860
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5097262631955766e-05,
+      "loss": 1.0257,
+      "step": 64870
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5096500982527764e-05,
+      "loss": 1.0013,
+      "step": 64880
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.509573933309976e-05,
+      "loss": 1.0884,
+      "step": 64890
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5094977683671766e-05,
+      "loss": 0.9493,
+      "step": 64900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5094216034243756e-05,
+      "loss": 1.0456,
+      "step": 64910
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.509345438481576e-05,
+      "loss": 1.0513,
+      "step": 64920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.509269273538776e-05,
+      "loss": 1.1493,
+      "step": 64930
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5091931085959755e-05,
+      "loss": 1.1024,
+      "step": 64940
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.509116943653175e-05,
+      "loss": 1.0723,
+      "step": 64950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.509040778710376e-05,
+      "loss": 1.0373,
+      "step": 64960
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508964613767575e-05,
+      "loss": 1.0545,
+      "step": 64970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508888448824775e-05,
+      "loss": 1.1317,
+      "step": 64980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508812283881975e-05,
+      "loss": 1.0698,
+      "step": 64990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508736118939175e-05,
+      "loss": 0.997,
+      "step": 65000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5086599539963744e-05,
+      "loss": 1.0295,
+      "step": 65010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508583789053575e-05,
+      "loss": 1.0037,
+      "step": 65020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5085076241107746e-05,
+      "loss": 0.9501,
+      "step": 65030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5084314591679744e-05,
+      "loss": 1.1469,
+      "step": 65040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508355294225174e-05,
+      "loss": 1.0502,
+      "step": 65050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508279129282374e-05,
+      "loss": 1.0138,
+      "step": 65060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508202964339574e-05,
+      "loss": 1.1703,
+      "step": 65070
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508126799396774e-05,
+      "loss": 1.1294,
+      "step": 65080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.508050634453974e-05,
+      "loss": 1.1489,
+      "step": 65090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5079744695111735e-05,
+      "loss": 1.0478,
+      "step": 65100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.507898304568374e-05,
+      "loss": 1.0317,
+      "step": 65110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.507822139625573e-05,
+      "loss": 1.0354,
+      "step": 65120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5077459746827734e-05,
+      "loss": 1.195,
+      "step": 65130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.507669809739973e-05,
+      "loss": 1.1084,
+      "step": 65140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.507593644797173e-05,
+      "loss": 1.0317,
+      "step": 65150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.507517479854373e-05,
+      "loss": 1.083,
+      "step": 65160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.507441314911573e-05,
+      "loss": 0.9677,
+      "step": 65170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.507365149968772e-05,
+      "loss": 1.0501,
+      "step": 65180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.5072889850259726e-05,
+      "loss": 1.1023,
+      "step": 65190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.5072128200831723e-05,
+      "loss": 1.1871,
+      "step": 65200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.507136655140372e-05,
+      "loss": 0.9927,
+      "step": 65210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.507060490197572e-05,
+      "loss": 1.0577,
+      "step": 65220
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506984325254772e-05,
+      "loss": 1.2344,
+      "step": 65230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506908160311972e-05,
+      "loss": 1.0187,
+      "step": 65240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506831995369172e-05,
+      "loss": 1.0518,
+      "step": 65250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.5067558304263715e-05,
+      "loss": 0.9864,
+      "step": 65260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506679665483571e-05,
+      "loss": 1.0547,
+      "step": 65270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506603500540772e-05,
+      "loss": 1.157,
+      "step": 65280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506527335597971e-05,
+      "loss": 1.1792,
+      "step": 65290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506451170655171e-05,
+      "loss": 1.0739,
+      "step": 65300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506375005712371e-05,
+      "loss": 1.1158,
+      "step": 65310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506298840769571e-05,
+      "loss": 1.094,
+      "step": 65320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.5062226758267704e-05,
+      "loss": 1.0587,
+      "step": 65330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.506146510883971e-05,
+      "loss": 1.1139,
+      "step": 65340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.50607034594117e-05,
+      "loss": 1.0671,
+      "step": 65350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.50599418099837e-05,
+      "loss": 1.0341,
+      "step": 65360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.50591801605557e-05,
+      "loss": 1.0945,
+      "step": 65370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.50584185111277e-05,
+      "loss": 1.077,
+      "step": 65380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5057656861699696e-05,
+      "loss": 1.0465,
+      "step": 65390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.50568952122717e-05,
+      "loss": 1.0731,
+      "step": 65400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.50561335628437e-05,
+      "loss": 1.1018,
+      "step": 65410
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5055371913415695e-05,
+      "loss": 1.0555,
+      "step": 65420
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.505461026398769e-05,
+      "loss": 1.0277,
+      "step": 65430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.505384861455969e-05,
+      "loss": 1.0116,
+      "step": 65440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5053086965131694e-05,
+      "loss": 1.1724,
+      "step": 65450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.505232531570369e-05,
+      "loss": 1.0655,
+      "step": 65460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.505156366627569e-05,
+      "loss": 1.0677,
+      "step": 65470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5050802016847686e-05,
+      "loss": 1.0516,
+      "step": 65480
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.505004036741969e-05,
+      "loss": 1.072,
+      "step": 65490
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504927871799168e-05,
+      "loss": 1.0751,
+      "step": 65500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5048517068563686e-05,
+      "loss": 1.0902,
+      "step": 65510
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504775541913568e-05,
+      "loss": 1.0691,
+      "step": 65520
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504699376970768e-05,
+      "loss": 1.0248,
+      "step": 65530
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504623212027968e-05,
+      "loss": 1.0304,
+      "step": 65540
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504547047085168e-05,
+      "loss": 1.0469,
+      "step": 65550
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504470882142367e-05,
+      "loss": 1.0958,
+      "step": 65560
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504394717199568e-05,
+      "loss": 1.0756,
+      "step": 65570
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5043185522567675e-05,
+      "loss": 0.9956,
+      "step": 65580
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.504242387313967e-05,
+      "loss": 1.1294,
+      "step": 65590
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.504166222371167e-05,
+      "loss": 1.0199,
+      "step": 65600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5040900574283674e-05,
+      "loss": 1.0175,
+      "step": 65610
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.504013892485567e-05,
+      "loss": 1.1498,
+      "step": 65620
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503937727542767e-05,
+      "loss": 1.1994,
+      "step": 65630
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5038615625999666e-05,
+      "loss": 1.0759,
+      "step": 65640
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5037853976571664e-05,
+      "loss": 1.0762,
+      "step": 65650
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503709232714367e-05,
+      "loss": 1.0723,
+      "step": 65660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5036330677715665e-05,
+      "loss": 1.2246,
+      "step": 65670
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503556902828766e-05,
+      "loss": 1.2222,
+      "step": 65680
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503480737885966e-05,
+      "loss": 1.2128,
+      "step": 65690
+    },
+    {
+      "epoch": 3.0,
+      "eval_cer": 0.6260877729683217,
+      "eval_em": 0.007379375591296121,
+      "eval_f1": 0.007379375591296121,
+      "eval_loss": 0.9553089141845703,
+      "eval_runtime": 1008.9207,
+      "eval_samples_per_second": 10.477,
+      "eval_steps_per_second": 1.31,
+      "eval_wer": 0.9926206244087039,
+      "step": 65699
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5034045729431665e-05,
+      "loss": 1.1053,
+      "step": 65700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5033284080003655e-05,
+      "loss": 0.9579,
+      "step": 65710
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503252243057566e-05,
+      "loss": 1.0463,
+      "step": 65720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503176078114766e-05,
+      "loss": 1.0043,
+      "step": 65730
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5030999131719654e-05,
+      "loss": 1.0889,
+      "step": 65740
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.503023748229165e-05,
+      "loss": 1.0386,
+      "step": 65750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.502947583286365e-05,
+      "loss": 1.0516,
+      "step": 65760
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.502871418343565e-05,
+      "loss": 1.039,
+      "step": 65770
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.502795253400765e-05,
+      "loss": 1.1559,
+      "step": 65780
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.502719088457965e-05,
+      "loss": 0.9634,
+      "step": 65790
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5026429235151646e-05,
+      "loss": 1.1542,
+      "step": 65800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5025667585723643e-05,
+      "loss": 1.0391,
+      "step": 65810
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.502490593629564e-05,
+      "loss": 1.008,
+      "step": 65820
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5024144286867645e-05,
+      "loss": 1.1762,
+      "step": 65830
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.502338263743964e-05,
+      "loss": 1.0252,
+      "step": 65840
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.502262098801164e-05,
+      "loss": 0.9389,
+      "step": 65850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.502185933858364e-05,
+      "loss": 1.0525,
+      "step": 65860
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.502109768915564e-05,
+      "loss": 1.0276,
+      "step": 65870
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.502033603972763e-05,
+      "loss": 0.9529,
+      "step": 65880
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501957439029964e-05,
+      "loss": 1.0768,
+      "step": 65890
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5018812740871634e-05,
+      "loss": 0.9485,
+      "step": 65900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501805109144363e-05,
+      "loss": 1.1325,
+      "step": 65910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501728944201563e-05,
+      "loss": 0.9334,
+      "step": 65920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5016527792587633e-05,
+      "loss": 0.9356,
+      "step": 65930
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5015766143159624e-05,
+      "loss": 1.0426,
+      "step": 65940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501500449373163e-05,
+      "loss": 0.9113,
+      "step": 65950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5014242844303626e-05,
+      "loss": 0.9874,
+      "step": 65960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501348119487562e-05,
+      "loss": 1.1256,
+      "step": 65970
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501271954544762e-05,
+      "loss": 1.1482,
+      "step": 65980
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.5011957896019625e-05,
+      "loss": 1.1353,
+      "step": 65990
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501119624659162e-05,
+      "loss": 0.9221,
+      "step": 66000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.501043459716362e-05,
+      "loss": 1.0469,
+      "step": 66010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.500967294773562e-05,
+      "loss": 0.9969,
+      "step": 66020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.5008911298307615e-05,
+      "loss": 1.0478,
+      "step": 66030
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.500814964887962e-05,
+      "loss": 1.0545,
+      "step": 66040
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.5007387999451617e-05,
+      "loss": 1.0184,
+      "step": 66050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.5006626350023614e-05,
+      "loss": 1.0777,
+      "step": 66060
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.500586470059561e-05,
+      "loss": 1.068,
+      "step": 66070
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.5005103051167616e-05,
+      "loss": 1.1402,
+      "step": 66080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.5004341401739606e-05,
+      "loss": 1.1461,
+      "step": 66090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.500357975231161e-05,
+      "loss": 0.9999,
+      "step": 66100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.500281810288361e-05,
+      "loss": 0.9641,
+      "step": 66110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.5002056453455606e-05,
+      "loss": 1.1313,
+      "step": 66120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.50012948040276e-05,
+      "loss": 1.1023,
+      "step": 66130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.500053315459961e-05,
+      "loss": 0.9972,
+      "step": 66140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.49997715051716e-05,
+      "loss": 1.035,
+      "step": 66150
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.49990098557436e-05,
+      "loss": 0.9887,
+      "step": 66160
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.49982482063156e-05,
+      "loss": 1.1136,
+      "step": 66170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.49974865568876e-05,
+      "loss": 0.9667,
+      "step": 66180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.4996724907459595e-05,
+      "loss": 0.9664,
+      "step": 66190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.499596325803159e-05,
+      "loss": 1.0229,
+      "step": 66200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.4995201608603596e-05,
+      "loss": 0.9814,
+      "step": 66210
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.4994439959175594e-05,
+      "loss": 1.0423,
+      "step": 66220
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.499367830974759e-05,
+      "loss": 1.0561,
+      "step": 66230
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.499291666031959e-05,
+      "loss": 0.9709,
+      "step": 66240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.499215501089159e-05,
+      "loss": 1.0409,
+      "step": 66250
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4991393361463584e-05,
+      "loss": 1.0798,
+      "step": 66260
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.499063171203559e-05,
+      "loss": 1.0327,
+      "step": 66270
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4989870062607585e-05,
+      "loss": 1.0196,
+      "step": 66280
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498910841317958e-05,
+      "loss": 1.0107,
+      "step": 66290
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498834676375158e-05,
+      "loss": 0.9941,
+      "step": 66300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4987585114323585e-05,
+      "loss": 0.9652,
+      "step": 66310
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4986823464895575e-05,
+      "loss": 1.0065,
+      "step": 66320
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498606181546758e-05,
+      "loss": 1.0109,
+      "step": 66330
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498530016603958e-05,
+      "loss": 1.1052,
+      "step": 66340
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4984538516611574e-05,
+      "loss": 0.9169,
+      "step": 66350
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498377686718357e-05,
+      "loss": 1.0081,
+      "step": 66360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4983015217755576e-05,
+      "loss": 1.0657,
+      "step": 66370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4982253568327574e-05,
+      "loss": 0.9558,
+      "step": 66380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498149191889957e-05,
+      "loss": 1.0463,
+      "step": 66390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.498073026947157e-05,
+      "loss": 1.0799,
+      "step": 66400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4979968620043566e-05,
+      "loss": 1.1691,
+      "step": 66410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.497920697061557e-05,
+      "loss": 1.0672,
+      "step": 66420
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.497844532118757e-05,
+      "loss": 1.0241,
+      "step": 66430
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.4977683671759565e-05,
+      "loss": 1.0898,
+      "step": 66440
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.497692202233156e-05,
+      "loss": 1.1228,
+      "step": 66450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.497616037290357e-05,
+      "loss": 0.9988,
+      "step": 66460
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.497539872347556e-05,
+      "loss": 0.9657,
+      "step": 66470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.497463707404756e-05,
+      "loss": 1.0921,
+      "step": 66480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.497387542461956e-05,
+      "loss": 1.0865,
+      "step": 66490
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.497311377519156e-05,
+      "loss": 1.0381,
+      "step": 66500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4972352125763554e-05,
+      "loss": 1.0297,
+      "step": 66510
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.497159047633556e-05,
+      "loss": 0.9254,
+      "step": 66520
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.497082882690755e-05,
+      "loss": 1.1958,
+      "step": 66530
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4970067177479553e-05,
+      "loss": 1.0396,
+      "step": 66540
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496930552805155e-05,
+      "loss": 0.9905,
+      "step": 66550
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496854387862355e-05,
+      "loss": 1.1169,
+      "step": 66560
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4967782229195546e-05,
+      "loss": 0.9743,
+      "step": 66570
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496702057976755e-05,
+      "loss": 1.0625,
+      "step": 66580
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496625893033955e-05,
+      "loss": 1.169,
+      "step": 66590
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4965497280911545e-05,
+      "loss": 1.0,
+      "step": 66600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496473563148354e-05,
+      "loss": 0.9724,
+      "step": 66610
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496397398205554e-05,
+      "loss": 1.0257,
+      "step": 66620
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4963212332627544e-05,
+      "loss": 1.038,
+      "step": 66630
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496245068319954e-05,
+      "loss": 0.9941,
+      "step": 66640
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.496168903377154e-05,
+      "loss": 1.0938,
+      "step": 66650
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4960927384343537e-05,
+      "loss": 1.0732,
+      "step": 66660
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.4960165734915534e-05,
+      "loss": 1.0385,
+      "step": 66670
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.495940408548753e-05,
+      "loss": 0.9744,
+      "step": 66680
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4958642436059536e-05,
+      "loss": 1.0492,
+      "step": 66690
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4957880786631526e-05,
+      "loss": 1.0183,
+      "step": 66700
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495711913720353e-05,
+      "loss": 0.955,
+      "step": 66710
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495635748777553e-05,
+      "loss": 0.8853,
+      "step": 66720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4955595838347526e-05,
+      "loss": 0.9041,
+      "step": 66730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495483418891952e-05,
+      "loss": 1.0591,
+      "step": 66740
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495407253949153e-05,
+      "loss": 1.065,
+      "step": 66750
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495331089006352e-05,
+      "loss": 1.0799,
+      "step": 66760
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495254924063552e-05,
+      "loss": 1.0605,
+      "step": 66770
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495178759120752e-05,
+      "loss": 0.9751,
+      "step": 66780
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495102594177952e-05,
+      "loss": 0.9825,
+      "step": 66790
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.495026429235152e-05,
+      "loss": 0.9694,
+      "step": 66800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494950264292352e-05,
+      "loss": 1.0409,
+      "step": 66810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4948740993495516e-05,
+      "loss": 1.1026,
+      "step": 66820
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4947979344067514e-05,
+      "loss": 1.0291,
+      "step": 66830
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494721769463952e-05,
+      "loss": 1.0135,
+      "step": 66840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494645604521151e-05,
+      "loss": 1.0738,
+      "step": 66850
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494569439578351e-05,
+      "loss": 1.1364,
+      "step": 66860
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494493274635551e-05,
+      "loss": 1.0549,
+      "step": 66870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494417109692751e-05,
+      "loss": 0.9595,
+      "step": 66880
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.4943409447499505e-05,
+      "loss": 1.1151,
+      "step": 66890
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.494264779807151e-05,
+      "loss": 1.023,
+      "step": 66900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.49418861486435e-05,
+      "loss": 1.1719,
+      "step": 66910
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4941124499215505e-05,
+      "loss": 1.118,
+      "step": 66920
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.49403628497875e-05,
+      "loss": 1.0757,
+      "step": 66930
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.49396012003595e-05,
+      "loss": 1.0342,
+      "step": 66940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.49388395509315e-05,
+      "loss": 0.972,
+      "step": 66950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.49380779015035e-05,
+      "loss": 1.2263,
+      "step": 66960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.49373162520755e-05,
+      "loss": 1.0281,
+      "step": 66970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4936554602647496e-05,
+      "loss": 1.1089,
+      "step": 66980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4935792953219494e-05,
+      "loss": 1.1886,
+      "step": 66990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.493503130379149e-05,
+      "loss": 1.1032,
+      "step": 67000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4934269654363495e-05,
+      "loss": 1.0448,
+      "step": 67010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.493350800493549e-05,
+      "loss": 1.0293,
+      "step": 67020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.493274635550749e-05,
+      "loss": 1.0663,
+      "step": 67030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.493198470607949e-05,
+      "loss": 1.2095,
+      "step": 67040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.493122305665149e-05,
+      "loss": 1.0144,
+      "step": 67050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.493046140722348e-05,
+      "loss": 1.0147,
+      "step": 67060
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.492969975779549e-05,
+      "loss": 1.0612,
+      "step": 67070
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4928938108367484e-05,
+      "loss": 0.9389,
+      "step": 67080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.492817645893948e-05,
+      "loss": 1.0885,
+      "step": 67090
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.492741480951148e-05,
+      "loss": 1.1569,
+      "step": 67100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4926653160083484e-05,
+      "loss": 0.9974,
+      "step": 67110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.4925891510655474e-05,
+      "loss": 1.0414,
+      "step": 67120
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.492512986122748e-05,
+      "loss": 1.0841,
+      "step": 67130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.492436821179947e-05,
+      "loss": 1.0715,
+      "step": 67140
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.4923606562371473e-05,
+      "loss": 1.0319,
+      "step": 67150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.492284491294347e-05,
+      "loss": 1.2046,
+      "step": 67160
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.492208326351547e-05,
+      "loss": 1.1357,
+      "step": 67170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.492132161408747e-05,
+      "loss": 1.0783,
+      "step": 67180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.492055996465947e-05,
+      "loss": 1.0421,
+      "step": 67190
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491979831523147e-05,
+      "loss": 1.048,
+      "step": 67200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.4919036665803465e-05,
+      "loss": 0.9272,
+      "step": 67210
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491827501637547e-05,
+      "loss": 1.0837,
+      "step": 67220
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491751336694746e-05,
+      "loss": 1.0318,
+      "step": 67230
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.4916751717519464e-05,
+      "loss": 0.9191,
+      "step": 67240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491599006809146e-05,
+      "loss": 1.0248,
+      "step": 67250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491522841866346e-05,
+      "loss": 1.0908,
+      "step": 67260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.4914466769235457e-05,
+      "loss": 0.9501,
+      "step": 67270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491370511980746e-05,
+      "loss": 1.0746,
+      "step": 67280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491294347037945e-05,
+      "loss": 1.0412,
+      "step": 67290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.4912181820951456e-05,
+      "loss": 1.0959,
+      "step": 67300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491142017152345e-05,
+      "loss": 0.9948,
+      "step": 67310
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.491065852209545e-05,
+      "loss": 1.1691,
+      "step": 67320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.490989687266745e-05,
+      "loss": 0.9809,
+      "step": 67330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.490913522323945e-05,
+      "loss": 1.1083,
+      "step": 67340
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490837357381144e-05,
+      "loss": 1.0573,
+      "step": 67350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490761192438345e-05,
+      "loss": 0.9914,
+      "step": 67360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4906850274955445e-05,
+      "loss": 1.014,
+      "step": 67370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490608862552744e-05,
+      "loss": 1.0594,
+      "step": 67380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4905326976099447e-05,
+      "loss": 1.0767,
+      "step": 67390
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4904565326671444e-05,
+      "loss": 1.0718,
+      "step": 67400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490380367724344e-05,
+      "loss": 1.0435,
+      "step": 67410
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490304202781544e-05,
+      "loss": 1.0751,
+      "step": 67420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490228037838744e-05,
+      "loss": 1.146,
+      "step": 67430
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4901518728959434e-05,
+      "loss": 1.0654,
+      "step": 67440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.490075707953144e-05,
+      "loss": 0.9504,
+      "step": 67450
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4899995430103436e-05,
+      "loss": 1.165,
+      "step": 67460
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.489923378067543e-05,
+      "loss": 1.0601,
+      "step": 67470
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.489847213124743e-05,
+      "loss": 1.1628,
+      "step": 67480
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4897710481819435e-05,
+      "loss": 1.0831,
+      "step": 67490
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4896948832391425e-05,
+      "loss": 1.1003,
+      "step": 67500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.489618718296343e-05,
+      "loss": 1.0731,
+      "step": 67510
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.489542553353543e-05,
+      "loss": 1.0665,
+      "step": 67520
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4894663884107425e-05,
+      "loss": 1.1607,
+      "step": 67530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.489390223467942e-05,
+      "loss": 1.1477,
+      "step": 67540
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.4893140585251426e-05,
+      "loss": 1.1705,
+      "step": 67550
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.489237893582342e-05,
+      "loss": 1.1035,
+      "step": 67560
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.489161728639542e-05,
+      "loss": 1.0838,
+      "step": 67570
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.489085563696742e-05,
+      "loss": 1.0293,
+      "step": 67580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4890093987539416e-05,
+      "loss": 1.1297,
+      "step": 67590
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488933233811142e-05,
+      "loss": 1.1413,
+      "step": 67600
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488857068868341e-05,
+      "loss": 1.0693,
+      "step": 67610
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4887809039255415e-05,
+      "loss": 0.9367,
+      "step": 67620
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488704738982741e-05,
+      "loss": 1.0564,
+      "step": 67630
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488628574039941e-05,
+      "loss": 1.124,
+      "step": 67640
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488552409097141e-05,
+      "loss": 1.088,
+      "step": 67650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488476244154341e-05,
+      "loss": 1.2148,
+      "step": 67660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.48840007921154e-05,
+      "loss": 1.0606,
+      "step": 67670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.488323914268741e-05,
+      "loss": 1.132,
+      "step": 67680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4882477493259404e-05,
+      "loss": 1.0205,
+      "step": 67690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.48817158438314e-05,
+      "loss": 1.0391,
+      "step": 67700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.48809541944034e-05,
+      "loss": 1.0907,
+      "step": 67710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4880192544975404e-05,
+      "loss": 1.0572,
+      "step": 67720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4879430895547394e-05,
+      "loss": 0.9717,
+      "step": 67730
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.48786692461194e-05,
+      "loss": 0.9812,
+      "step": 67740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4877907596691396e-05,
+      "loss": 1.1797,
+      "step": 67750
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.4877145947263393e-05,
+      "loss": 1.018,
+      "step": 67760
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.48763842978354e-05,
+      "loss": 1.0719,
+      "step": 67770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4875622648407395e-05,
+      "loss": 1.0461,
+      "step": 67780
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.487486099897939e-05,
+      "loss": 0.9739,
+      "step": 67790
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.487409934955139e-05,
+      "loss": 1.0392,
+      "step": 67800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4873337700123394e-05,
+      "loss": 1.1473,
+      "step": 67810
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4872576050695385e-05,
+      "loss": 1.1484,
+      "step": 67820
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.487181440126739e-05,
+      "loss": 0.978,
+      "step": 67830
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.487105275183939e-05,
+      "loss": 1.0898,
+      "step": 67840
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4870291102411384e-05,
+      "loss": 0.9962,
+      "step": 67850
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486952945298338e-05,
+      "loss": 0.9742,
+      "step": 67860
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4868767803555386e-05,
+      "loss": 1.1655,
+      "step": 67870
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4868006154127377e-05,
+      "loss": 1.1531,
+      "step": 67880
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486724450469938e-05,
+      "loss": 1.0354,
+      "step": 67890
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486648285527138e-05,
+      "loss": 1.2046,
+      "step": 67900
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.4865721205843376e-05,
+      "loss": 1.1507,
+      "step": 67910
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486495955641537e-05,
+      "loss": 1.0205,
+      "step": 67920
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486419790698738e-05,
+      "loss": 1.1123,
+      "step": 67930
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486343625755937e-05,
+      "loss": 1.0978,
+      "step": 67940
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486267460813137e-05,
+      "loss": 1.0192,
+      "step": 67950
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486191295870337e-05,
+      "loss": 1.106,
+      "step": 67960
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486115130927537e-05,
+      "loss": 1.1387,
+      "step": 67970
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.486038965984737e-05,
+      "loss": 1.0787,
+      "step": 67980
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.485962801041937e-05,
+      "loss": 0.9267,
+      "step": 67990
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4858866360991367e-05,
+      "loss": 1.1822,
+      "step": 68000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4858104711563364e-05,
+      "loss": 1.1149,
+      "step": 68010
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.485734306213537e-05,
+      "loss": 1.0241,
+      "step": 68020
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.485658141270736e-05,
+      "loss": 0.9774,
+      "step": 68030
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.485581976327936e-05,
+      "loss": 1.0208,
+      "step": 68040
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4855058113851354e-05,
+      "loss": 1.0972,
+      "step": 68050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.485429646442336e-05,
+      "loss": 0.9955,
+      "step": 68060
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4853534814995356e-05,
+      "loss": 1.0981,
+      "step": 68070
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.485277316556735e-05,
+      "loss": 1.0561,
+      "step": 68080
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.485201151613935e-05,
+      "loss": 1.011,
+      "step": 68090
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4851249866711355e-05,
+      "loss": 1.1824,
+      "step": 68100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4850488217283345e-05,
+      "loss": 0.9961,
+      "step": 68110
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.484972656785535e-05,
+      "loss": 1.1125,
+      "step": 68120
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.484896491842735e-05,
+      "loss": 1.0343,
+      "step": 68130
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4848203268999345e-05,
+      "loss": 1.2574,
+      "step": 68140
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.484744161957134e-05,
+      "loss": 1.0256,
+      "step": 68150
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4846679970143346e-05,
+      "loss": 1.0328,
+      "step": 68160
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4845918320715344e-05,
+      "loss": 1.114,
+      "step": 68170
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.484515667128734e-05,
+      "loss": 1.003,
+      "step": 68180
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4844395021859346e-05,
+      "loss": 1.0417,
+      "step": 68190
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.4843633372431336e-05,
+      "loss": 1.1339,
+      "step": 68200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.484287172300334e-05,
+      "loss": 1.1572,
+      "step": 68210
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.484211007357534e-05,
+      "loss": 1.0899,
+      "step": 68220
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4841348424147335e-05,
+      "loss": 1.1445,
+      "step": 68230
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.484058677471933e-05,
+      "loss": 1.1758,
+      "step": 68240
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483982512529134e-05,
+      "loss": 1.0291,
+      "step": 68250
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483906347586333e-05,
+      "loss": 0.9801,
+      "step": 68260
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483830182643533e-05,
+      "loss": 1.0209,
+      "step": 68270
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483754017700733e-05,
+      "loss": 1.1793,
+      "step": 68280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483677852757933e-05,
+      "loss": 1.0892,
+      "step": 68290
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4836016878151324e-05,
+      "loss": 0.9826,
+      "step": 68300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483525522872333e-05,
+      "loss": 1.0549,
+      "step": 68310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483449357929532e-05,
+      "loss": 1.0326,
+      "step": 68320
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4833731929867324e-05,
+      "loss": 0.9286,
+      "step": 68330
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483297028043932e-05,
+      "loss": 1.0219,
+      "step": 68340
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483220863101132e-05,
+      "loss": 1.002,
+      "step": 68350
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4831446981583316e-05,
+      "loss": 1.0339,
+      "step": 68360
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.483068533215532e-05,
+      "loss": 0.9658,
+      "step": 68370
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.482992368272732e-05,
+      "loss": 1.1491,
+      "step": 68380
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4829162033299315e-05,
+      "loss": 1.1553,
+      "step": 68390
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.482840038387132e-05,
+      "loss": 1.1222,
+      "step": 68400
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.482763873444331e-05,
+      "loss": 1.1664,
+      "step": 68410
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4826877085015314e-05,
+      "loss": 1.0295,
+      "step": 68420
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.482611543558731e-05,
+      "loss": 1.0867,
+      "step": 68430
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.482535378615931e-05,
+      "loss": 1.0768,
+      "step": 68440
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.482459213673131e-05,
+      "loss": 1.0606,
+      "step": 68450
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.482383048730331e-05,
+      "loss": 1.1382,
+      "step": 68460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.48230688378753e-05,
+      "loss": 1.0501,
+      "step": 68470
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.4822307188447306e-05,
+      "loss": 1.143,
+      "step": 68480
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.48215455390193e-05,
+      "loss": 1.0085,
+      "step": 68490
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.48207838895913e-05,
+      "loss": 0.9586,
+      "step": 68500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.48200222401633e-05,
+      "loss": 0.9838,
+      "step": 68510
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.4819260590735296e-05,
+      "loss": 1.0809,
+      "step": 68520
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481849894130729e-05,
+      "loss": 1.0155,
+      "step": 68530
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.48177372918793e-05,
+      "loss": 1.0035,
+      "step": 68540
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.4816975642451295e-05,
+      "loss": 1.0994,
+      "step": 68550
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481621399302329e-05,
+      "loss": 0.9983,
+      "step": 68560
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.48154523435953e-05,
+      "loss": 1.1656,
+      "step": 68570
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481469069416729e-05,
+      "loss": 0.9205,
+      "step": 68580
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481392904473929e-05,
+      "loss": 0.9447,
+      "step": 68590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481316739531129e-05,
+      "loss": 0.9303,
+      "step": 68600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.4812405745883287e-05,
+      "loss": 1.1328,
+      "step": 68610
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.4811644096455284e-05,
+      "loss": 1.0873,
+      "step": 68620
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481088244702729e-05,
+      "loss": 0.9384,
+      "step": 68630
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.481012079759928e-05,
+      "loss": 1.0292,
+      "step": 68640
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.480935914817128e-05,
+      "loss": 1.1593,
+      "step": 68650
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480859749874328e-05,
+      "loss": 1.0026,
+      "step": 68660
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480783584931528e-05,
+      "loss": 1.0878,
+      "step": 68670
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.4807074199887276e-05,
+      "loss": 0.9473,
+      "step": 68680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480631255045928e-05,
+      "loss": 0.9646,
+      "step": 68690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480555090103127e-05,
+      "loss": 1.1539,
+      "step": 68700
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.4804789251603275e-05,
+      "loss": 0.976,
+      "step": 68710
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480402760217527e-05,
+      "loss": 0.9379,
+      "step": 68720
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480326595274727e-05,
+      "loss": 1.0626,
+      "step": 68730
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480250430331927e-05,
+      "loss": 1.1423,
+      "step": 68740
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480174265389127e-05,
+      "loss": 1.0447,
+      "step": 68750
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.480098100446327e-05,
+      "loss": 1.1227,
+      "step": 68760
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.4800219355035266e-05,
+      "loss": 0.9847,
+      "step": 68770
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479945770560727e-05,
+      "loss": 0.9764,
+      "step": 68780
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479869605617926e-05,
+      "loss": 0.9932,
+      "step": 68790
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.4797934406751266e-05,
+      "loss": 1.0508,
+      "step": 68800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479717275732326e-05,
+      "loss": 1.1373,
+      "step": 68810
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479641110789526e-05,
+      "loss": 1.0193,
+      "step": 68820
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479564945846726e-05,
+      "loss": 1.0096,
+      "step": 68830
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479488780903926e-05,
+      "loss": 0.9222,
+      "step": 68840
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479412615961125e-05,
+      "loss": 0.9175,
+      "step": 68850
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.479336451018326e-05,
+      "loss": 0.9328,
+      "step": 68860
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.4792602860755255e-05,
+      "loss": 1.2473,
+      "step": 68870
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.479184121132725e-05,
+      "loss": 1.0178,
+      "step": 68880
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.479107956189925e-05,
+      "loss": 1.051,
+      "step": 68890
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4790317912471254e-05,
+      "loss": 1.0962,
+      "step": 68900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4789556263043244e-05,
+      "loss": 1.0395,
+      "step": 68910
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478879461361525e-05,
+      "loss": 1.1946,
+      "step": 68920
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4788032964187246e-05,
+      "loss": 0.9993,
+      "step": 68930
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4787271314759244e-05,
+      "loss": 0.9591,
+      "step": 68940
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478650966533124e-05,
+      "loss": 1.1289,
+      "step": 68950
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478574801590324e-05,
+      "loss": 1.0237,
+      "step": 68960
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478498636647524e-05,
+      "loss": 1.0244,
+      "step": 68970
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478422471704724e-05,
+      "loss": 1.0129,
+      "step": 68980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478346306761924e-05,
+      "loss": 1.0223,
+      "step": 68990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4782701418191235e-05,
+      "loss": 1.0028,
+      "step": 69000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478193976876324e-05,
+      "loss": 0.9705,
+      "step": 69010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.478117811933523e-05,
+      "loss": 1.0782,
+      "step": 69020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4780416469907234e-05,
+      "loss": 1.0788,
+      "step": 69030
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.477965482047923e-05,
+      "loss": 1.051,
+      "step": 69040
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.477889317105123e-05,
+      "loss": 1.0099,
+      "step": 69050
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.477813152162323e-05,
+      "loss": 1.0914,
+      "step": 69060
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.477736987219523e-05,
+      "loss": 0.995,
+      "step": 69070
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.477660822276722e-05,
+      "loss": 1.032,
+      "step": 69080
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.4775846573339226e-05,
+      "loss": 1.0415,
+      "step": 69090
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4775084923911223e-05,
+      "loss": 1.0347,
+      "step": 69100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.477432327448322e-05,
+      "loss": 1.0151,
+      "step": 69110
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.477356162505522e-05,
+      "loss": 0.9176,
+      "step": 69120
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.477279997562722e-05,
+      "loss": 0.9946,
+      "step": 69130
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.477203832619922e-05,
+      "loss": 1.0302,
+      "step": 69140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.477127667677122e-05,
+      "loss": 1.0814,
+      "step": 69150
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4770515027343215e-05,
+      "loss": 1.0045,
+      "step": 69160
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.476975337791521e-05,
+      "loss": 1.1038,
+      "step": 69170
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.476899172848722e-05,
+      "loss": 1.0174,
+      "step": 69180
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4768230079059214e-05,
+      "loss": 1.0423,
+      "step": 69190
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.476746842963121e-05,
+      "loss": 1.1341,
+      "step": 69200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.476670678020321e-05,
+      "loss": 1.0847,
+      "step": 69210
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.476594513077521e-05,
+      "loss": 0.9774,
+      "step": 69220
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4765183481347204e-05,
+      "loss": 1.1391,
+      "step": 69230
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.476442183191921e-05,
+      "loss": 0.9338,
+      "step": 69240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4763660182491206e-05,
+      "loss": 1.1094,
+      "step": 69250
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.47628985330632e-05,
+      "loss": 1.0267,
+      "step": 69260
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.47621368836352e-05,
+      "loss": 1.0291,
+      "step": 69270
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4761375234207205e-05,
+      "loss": 1.183,
+      "step": 69280
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.4760613584779196e-05,
+      "loss": 1.0879,
+      "step": 69290
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.47598519353512e-05,
+      "loss": 0.9648,
+      "step": 69300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.47590902859232e-05,
+      "loss": 1.0763,
+      "step": 69310
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4758328636495195e-05,
+      "loss": 1.0994,
+      "step": 69320
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.475756698706719e-05,
+      "loss": 1.0649,
+      "step": 69330
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4756805337639196e-05,
+      "loss": 0.9722,
+      "step": 69340
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4756043688211194e-05,
+      "loss": 1.0504,
+      "step": 69350
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.475528203878319e-05,
+      "loss": 1.1201,
+      "step": 69360
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4754520389355196e-05,
+      "loss": 0.876,
+      "step": 69370
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4753758739927186e-05,
+      "loss": 1.0679,
+      "step": 69380
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.475299709049919e-05,
+      "loss": 0.9483,
+      "step": 69390
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.475223544107119e-05,
+      "loss": 1.0652,
+      "step": 69400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4751473791643186e-05,
+      "loss": 1.0303,
+      "step": 69410
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.475071214221518e-05,
+      "loss": 0.9116,
+      "step": 69420
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474995049278718e-05,
+      "loss": 1.0173,
+      "step": 69430
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474918884335918e-05,
+      "loss": 0.9382,
+      "step": 69440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474842719393118e-05,
+      "loss": 0.9821,
+      "step": 69450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474766554450317e-05,
+      "loss": 1.1049,
+      "step": 69460
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474690389507518e-05,
+      "loss": 0.9875,
+      "step": 69470
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4746142245647175e-05,
+      "loss": 0.9613,
+      "step": 69480
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474538059621917e-05,
+      "loss": 1.1876,
+      "step": 69490
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474461894679117e-05,
+      "loss": 0.9516,
+      "step": 69500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.4743857297363174e-05,
+      "loss": 1.0993,
+      "step": 69510
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474309564793517e-05,
+      "loss": 1.125,
+      "step": 69520
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474233399850717e-05,
+      "loss": 1.0831,
+      "step": 69530
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4741572349079166e-05,
+      "loss": 1.0871,
+      "step": 69540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4740810699651164e-05,
+      "loss": 0.9789,
+      "step": 69550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.474004905022317e-05,
+      "loss": 0.9734,
+      "step": 69560
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4739287400795165e-05,
+      "loss": 0.9897,
+      "step": 69570
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473852575136716e-05,
+      "loss": 1.0643,
+      "step": 69580
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473776410193916e-05,
+      "loss": 1.0588,
+      "step": 69590
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4737002452511164e-05,
+      "loss": 1.0711,
+      "step": 69600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4736240803083155e-05,
+      "loss": 1.0767,
+      "step": 69610
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473547915365516e-05,
+      "loss": 1.0411,
+      "step": 69620
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473471750422716e-05,
+      "loss": 1.0495,
+      "step": 69630
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4733955854799154e-05,
+      "loss": 0.9821,
+      "step": 69640
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473319420537115e-05,
+      "loss": 0.9802,
+      "step": 69650
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4732432555943156e-05,
+      "loss": 1.0156,
+      "step": 69660
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473167090651515e-05,
+      "loss": 1.023,
+      "step": 69670
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473090925708715e-05,
+      "loss": 1.0906,
+      "step": 69680
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.473014760765915e-05,
+      "loss": 1.0224,
+      "step": 69690
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4729385958231146e-05,
+      "loss": 1.0007,
+      "step": 69700
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4728624308803143e-05,
+      "loss": 0.9842,
+      "step": 69710
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.472786265937515e-05,
+      "loss": 1.0534,
+      "step": 69720
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.4727101009947145e-05,
+      "loss": 1.0739,
+      "step": 69730
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.472633936051914e-05,
+      "loss": 1.0436,
+      "step": 69740
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.472557771109114e-05,
+      "loss": 1.1384,
+      "step": 69750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.472481606166314e-05,
+      "loss": 0.9693,
+      "step": 69760
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.472405441223514e-05,
+      "loss": 1.0525,
+      "step": 69770
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.472329276280714e-05,
+      "loss": 1.1444,
+      "step": 69780
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.472253111337914e-05,
+      "loss": 0.9474,
+      "step": 69790
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.4721769463951134e-05,
+      "loss": 1.1092,
+      "step": 69800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.472100781452314e-05,
+      "loss": 0.9828,
+      "step": 69810
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.472024616509513e-05,
+      "loss": 1.0714,
+      "step": 69820
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471948451566713e-05,
+      "loss": 1.0783,
+      "step": 69830
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471872286623913e-05,
+      "loss": 0.9891,
+      "step": 69840
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471796121681113e-05,
+      "loss": 0.9896,
+      "step": 69850
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.4717199567383126e-05,
+      "loss": 1.1328,
+      "step": 69860
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471643791795513e-05,
+      "loss": 1.172,
+      "step": 69870
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471567626852712e-05,
+      "loss": 1.0405,
+      "step": 69880
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.4714914619099125e-05,
+      "loss": 1.1491,
+      "step": 69890
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471415296967112e-05,
+      "loss": 1.2362,
+      "step": 69900
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471339132024312e-05,
+      "loss": 0.9643,
+      "step": 69910
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471262967081512e-05,
+      "loss": 1.0777,
+      "step": 69920
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.4711868021387115e-05,
+      "loss": 0.9693,
+      "step": 69930
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.471110637195912e-05,
+      "loss": 1.0721,
+      "step": 69940
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.4710344722531116e-05,
+      "loss": 1.0438,
+      "step": 69950
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.4709583073103114e-05,
+      "loss": 1.1583,
+      "step": 69960
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.470882142367511e-05,
+      "loss": 1.1308,
+      "step": 69970
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4708059774247116e-05,
+      "loss": 1.0852,
+      "step": 69980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4707298124819106e-05,
+      "loss": 1.1852,
+      "step": 69990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.470653647539111e-05,
+      "loss": 1.0005,
+      "step": 70000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.470577482596311e-05,
+      "loss": 1.0814,
+      "step": 70010
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4705013176535106e-05,
+      "loss": 1.0378,
+      "step": 70020
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.47042515271071e-05,
+      "loss": 1.0684,
+      "step": 70030
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.470348987767911e-05,
+      "loss": 1.0175,
+      "step": 70040
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.47027282282511e-05,
+      "loss": 0.9903,
+      "step": 70050
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.47019665788231e-05,
+      "loss": 1.0861,
+      "step": 70060
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.47012049293951e-05,
+      "loss": 1.101,
+      "step": 70070
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.47004432799671e-05,
+      "loss": 0.9985,
+      "step": 70080
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4699681630539095e-05,
+      "loss": 1.0132,
+      "step": 70090
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.46989199811111e-05,
+      "loss": 1.1111,
+      "step": 70100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4698158331683096e-05,
+      "loss": 1.1,
+      "step": 70110
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4697396682255094e-05,
+      "loss": 1.0587,
+      "step": 70120
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.469663503282709e-05,
+      "loss": 1.0883,
+      "step": 70130
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.469587338339909e-05,
+      "loss": 1.1438,
+      "step": 70140
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.469511173397109e-05,
+      "loss": 1.0499,
+      "step": 70150
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.469435008454309e-05,
+      "loss": 0.9476,
+      "step": 70160
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.469358843511509e-05,
+      "loss": 1.0956,
+      "step": 70170
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4692826785687085e-05,
+      "loss": 1.1378,
+      "step": 70180
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.469206513625909e-05,
+      "loss": 1.0782,
+      "step": 70190
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.469130348683108e-05,
+      "loss": 0.8913,
+      "step": 70200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.4690541837403085e-05,
+      "loss": 1.2159,
+      "step": 70210
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468978018797508e-05,
+      "loss": 1.096,
+      "step": 70220
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468901853854708e-05,
+      "loss": 1.0537,
+      "step": 70230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468825688911908e-05,
+      "loss": 1.0182,
+      "step": 70240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468749523969108e-05,
+      "loss": 0.9846,
+      "step": 70250
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468673359026307e-05,
+      "loss": 1.0911,
+      "step": 70260
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.4685971940835076e-05,
+      "loss": 1.0379,
+      "step": 70270
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.4685210291407074e-05,
+      "loss": 1.0146,
+      "step": 70280
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468444864197907e-05,
+      "loss": 1.1818,
+      "step": 70290
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468368699255107e-05,
+      "loss": 1.099,
+      "step": 70300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468292534312307e-05,
+      "loss": 1.0266,
+      "step": 70310
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468216369369507e-05,
+      "loss": 1.1407,
+      "step": 70320
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.468140204426707e-05,
+      "loss": 1.1403,
+      "step": 70330
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.4680640394839065e-05,
+      "loss": 1.0401,
+      "step": 70340
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.467987874541106e-05,
+      "loss": 0.9805,
+      "step": 70350
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.467911709598307e-05,
+      "loss": 1.1966,
+      "step": 70360
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.467835544655506e-05,
+      "loss": 1.0959,
+      "step": 70370
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.467759379712706e-05,
+      "loss": 1.1542,
+      "step": 70380
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.467683214769906e-05,
+      "loss": 0.9883,
+      "step": 70390
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.467607049827106e-05,
+      "loss": 1.0033,
+      "step": 70400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.4675308848843054e-05,
+      "loss": 1.0276,
+      "step": 70410
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.467454719941506e-05,
+      "loss": 1.0596,
+      "step": 70420
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.467378554998705e-05,
+      "loss": 0.9711,
+      "step": 70430
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.467302390055905e-05,
+      "loss": 0.9968,
+      "step": 70440
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.467226225113105e-05,
+      "loss": 0.9564,
+      "step": 70450
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.467150060170305e-05,
+      "loss": 1.0634,
+      "step": 70460
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.4670738952275046e-05,
+      "loss": 0.9924,
+      "step": 70470
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466997730284705e-05,
+      "loss": 1.013,
+      "step": 70480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466921565341905e-05,
+      "loss": 1.0669,
+      "step": 70490
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.4668454003991045e-05,
+      "loss": 0.9322,
+      "step": 70500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466769235456304e-05,
+      "loss": 0.9289,
+      "step": 70510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466693070513504e-05,
+      "loss": 0.9915,
+      "step": 70520
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.4666169055707044e-05,
+      "loss": 1.1084,
+      "step": 70530
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466540740627904e-05,
+      "loss": 0.9902,
+      "step": 70540
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466464575685104e-05,
+      "loss": 0.9679,
+      "step": 70550
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.4663884107423036e-05,
+      "loss": 1.1052,
+      "step": 70560
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466312245799504e-05,
+      "loss": 1.016,
+      "step": 70570
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466236080856703e-05,
+      "loss": 1.004,
+      "step": 70580
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.4661599159139036e-05,
+      "loss": 1.0927,
+      "step": 70590
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466083750971103e-05,
+      "loss": 1.284,
+      "step": 70600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.466007586028303e-05,
+      "loss": 1.027,
+      "step": 70610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.465931421085503e-05,
+      "loss": 1.0656,
+      "step": 70620
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465855256142703e-05,
+      "loss": 0.9304,
+      "step": 70630
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465779091199902e-05,
+      "loss": 0.9822,
+      "step": 70640
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465702926257103e-05,
+      "loss": 1.0445,
+      "step": 70650
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4656267613143025e-05,
+      "loss": 0.9944,
+      "step": 70660
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465550596371502e-05,
+      "loss": 1.1504,
+      "step": 70670
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465474431428702e-05,
+      "loss": 0.9248,
+      "step": 70680
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4653982664859024e-05,
+      "loss": 0.9787,
+      "step": 70690
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465322101543102e-05,
+      "loss": 0.9771,
+      "step": 70700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465245936600302e-05,
+      "loss": 1.1146,
+      "step": 70710
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4651697716575016e-05,
+      "loss": 1.0158,
+      "step": 70720
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4650936067147014e-05,
+      "loss": 1.0365,
+      "step": 70730
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.465017441771902e-05,
+      "loss": 1.0553,
+      "step": 70740
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4649412768291015e-05,
+      "loss": 0.9934,
+      "step": 70750
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.464865111886301e-05,
+      "loss": 0.991,
+      "step": 70760
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.464788946943501e-05,
+      "loss": 0.9438,
+      "step": 70770
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4647127820007015e-05,
+      "loss": 1.0495,
+      "step": 70780
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4646366170579005e-05,
+      "loss": 1.0088,
+      "step": 70790
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.464560452115101e-05,
+      "loss": 1.0668,
+      "step": 70800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4644842871723e-05,
+      "loss": 1.1068,
+      "step": 70810
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4644081222295005e-05,
+      "loss": 0.9734,
+      "step": 70820
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4643319572867e-05,
+      "loss": 0.9572,
+      "step": 70830
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.4642557923439e-05,
+      "loss": 1.0993,
+      "step": 70840
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4641796274011e-05,
+      "loss": 1.1603,
+      "step": 70850
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4641034624583e-05,
+      "loss": 1.0229,
+      "step": 70860
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4640272975155e-05,
+      "loss": 1.1198,
+      "step": 70870
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4639511325726996e-05,
+      "loss": 1.1261,
+      "step": 70880
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4638749676298994e-05,
+      "loss": 1.1143,
+      "step": 70890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463798802687099e-05,
+      "loss": 0.9909,
+      "step": 70900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4637226377442995e-05,
+      "loss": 1.1998,
+      "step": 70910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463646472801499e-05,
+      "loss": 1.0799,
+      "step": 70920
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463570307858699e-05,
+      "loss": 0.955,
+      "step": 70930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463494142915899e-05,
+      "loss": 0.9354,
+      "step": 70940
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463417977973099e-05,
+      "loss": 1.0277,
+      "step": 70950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463341813030298e-05,
+      "loss": 1.1064,
+      "step": 70960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463265648087499e-05,
+      "loss": 1.0386,
+      "step": 70970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4631894831446984e-05,
+      "loss": 1.0286,
+      "step": 70980
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463113318201898e-05,
+      "loss": 0.9195,
+      "step": 70990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.463037153259098e-05,
+      "loss": 1.1714,
+      "step": 71000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4629609883162983e-05,
+      "loss": 1.0745,
+      "step": 71010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4628848233734974e-05,
+      "loss": 1.1808,
+      "step": 71020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.462808658430698e-05,
+      "loss": 1.0701,
+      "step": 71030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.4627324934878976e-05,
+      "loss": 1.0351,
+      "step": 71040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.462656328545097e-05,
+      "loss": 0.8624,
+      "step": 71050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.462580163602297e-05,
+      "loss": 1.0288,
+      "step": 71060
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4625039986594975e-05,
+      "loss": 0.9862,
+      "step": 71070
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.462427833716697e-05,
+      "loss": 1.0633,
+      "step": 71080
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.462351668773897e-05,
+      "loss": 1.0201,
+      "step": 71090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.462275503831097e-05,
+      "loss": 1.2157,
+      "step": 71100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4621993388882965e-05,
+      "loss": 1.1106,
+      "step": 71110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.462123173945497e-05,
+      "loss": 1.0429,
+      "step": 71120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.462047009002697e-05,
+      "loss": 0.9809,
+      "step": 71130
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4619708440598964e-05,
+      "loss": 1.0453,
+      "step": 71140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461894679117096e-05,
+      "loss": 1.1491,
+      "step": 71150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4618185141742966e-05,
+      "loss": 1.0589,
+      "step": 71160
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4617423492314957e-05,
+      "loss": 1.066,
+      "step": 71170
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461666184288696e-05,
+      "loss": 1.1343,
+      "step": 71180
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461590019345896e-05,
+      "loss": 0.972,
+      "step": 71190
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4615138544030956e-05,
+      "loss": 0.9862,
+      "step": 71200
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461437689460295e-05,
+      "loss": 0.9923,
+      "step": 71210
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461361524517496e-05,
+      "loss": 1.0314,
+      "step": 71220
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461285359574695e-05,
+      "loss": 1.0127,
+      "step": 71230
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461209194631895e-05,
+      "loss": 1.0344,
+      "step": 71240
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461133029689095e-05,
+      "loss": 0.9477,
+      "step": 71250
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.461056864746295e-05,
+      "loss": 0.9767,
+      "step": 71260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4609806998034945e-05,
+      "loss": 1.0389,
+      "step": 71270
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.460904534860694e-05,
+      "loss": 1.0095,
+      "step": 71280
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4608283699178946e-05,
+      "loss": 1.1322,
+      "step": 71290
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4607522049750944e-05,
+      "loss": 1.2625,
+      "step": 71300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.460676040032294e-05,
+      "loss": 1.0445,
+      "step": 71310
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.460599875089494e-05,
+      "loss": 1.0261,
+      "step": 71320
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.460523710146694e-05,
+      "loss": 0.9043,
+      "step": 71330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4604475452038934e-05,
+      "loss": 1.0962,
+      "step": 71340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.460371380261094e-05,
+      "loss": 1.0532,
+      "step": 71350
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4602952153182935e-05,
+      "loss": 1.0817,
+      "step": 71360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.460219050375493e-05,
+      "loss": 1.0484,
+      "step": 71370
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.460142885432693e-05,
+      "loss": 1.0478,
+      "step": 71380
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4600667204898935e-05,
+      "loss": 0.9366,
+      "step": 71390
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4599905555470925e-05,
+      "loss": 0.985,
+      "step": 71400
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459914390604293e-05,
+      "loss": 1.0154,
+      "step": 71410
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459838225661493e-05,
+      "loss": 1.0135,
+      "step": 71420
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4597620607186925e-05,
+      "loss": 0.9844,
+      "step": 71430
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459685895775892e-05,
+      "loss": 0.9817,
+      "step": 71440
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4596097308330926e-05,
+      "loss": 1.0111,
+      "step": 71450
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459533565890292e-05,
+      "loss": 1.0165,
+      "step": 71460
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459457400947492e-05,
+      "loss": 1.0858,
+      "step": 71470
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459381236004692e-05,
+      "loss": 1.045,
+      "step": 71480
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.4593050710618916e-05,
+      "loss": 0.9846,
+      "step": 71490
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.459228906119092e-05,
+      "loss": 1.0247,
+      "step": 71500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.459152741176292e-05,
+      "loss": 1.1366,
+      "step": 71510
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.4590765762334915e-05,
+      "loss": 1.1651,
+      "step": 71520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.459000411290691e-05,
+      "loss": 1.2316,
+      "step": 71530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.458924246347892e-05,
+      "loss": 1.0431,
+      "step": 71540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.458848081405091e-05,
+      "loss": 0.9986,
+      "step": 71550
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.458771916462291e-05,
+      "loss": 1.1331,
+      "step": 71560
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.458695751519491e-05,
+      "loss": 1.054,
+      "step": 71570
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.458619586576691e-05,
+      "loss": 1.0056,
+      "step": 71580
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.4585434216338904e-05,
+      "loss": 1.0139,
+      "step": 71590
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.458467256691091e-05,
+      "loss": 0.9571,
+      "step": 71600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.45839109174829e-05,
+      "loss": 1.0725,
+      "step": 71610
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.4583149268054904e-05,
+      "loss": 1.0045,
+      "step": 71620
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.45823876186269e-05,
+      "loss": 1.0524,
+      "step": 71630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.45816259691989e-05,
+      "loss": 1.075,
+      "step": 71640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.4580864319770896e-05,
+      "loss": 1.0649,
+      "step": 71650
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.45801026703429e-05,
+      "loss": 1.1136,
+      "step": 71660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.45793410209149e-05,
+      "loss": 1.0503,
+      "step": 71670
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.4578579371486895e-05,
+      "loss": 1.0701,
+      "step": 71680
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.457781772205889e-05,
+      "loss": 1.0831,
+      "step": 71690
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.457705607263089e-05,
+      "loss": 1.0196,
+      "step": 71700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.4576294423202894e-05,
+      "loss": 0.9698,
+      "step": 71710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.457553277377489e-05,
+      "loss": 1.1429,
+      "step": 71720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.457477112434689e-05,
+      "loss": 1.1223,
+      "step": 71730
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.457400947491889e-05,
+      "loss": 1.1142,
+      "step": 71740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4573247825490884e-05,
+      "loss": 1.0787,
+      "step": 71750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.457248617606288e-05,
+      "loss": 1.0311,
+      "step": 71760
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4571724526634886e-05,
+      "loss": 1.0883,
+      "step": 71770
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4570962877206877e-05,
+      "loss": 1.0582,
+      "step": 71780
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.457020122777888e-05,
+      "loss": 1.0554,
+      "step": 71790
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456943957835088e-05,
+      "loss": 1.0415,
+      "step": 71800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4568677928922876e-05,
+      "loss": 1.0049,
+      "step": 71810
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456791627949487e-05,
+      "loss": 1.0873,
+      "step": 71820
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456715463006688e-05,
+      "loss": 1.0804,
+      "step": 71830
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456639298063887e-05,
+      "loss": 1.0788,
+      "step": 71840
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456563133121087e-05,
+      "loss": 1.0622,
+      "step": 71850
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456486968178287e-05,
+      "loss": 1.1463,
+      "step": 71860
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456410803235487e-05,
+      "loss": 1.066,
+      "step": 71870
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456334638292687e-05,
+      "loss": 1.0338,
+      "step": 71880
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456258473349887e-05,
+      "loss": 1.2184,
+      "step": 71890
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4561823084070866e-05,
+      "loss": 1.1692,
+      "step": 71900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4561061434642864e-05,
+      "loss": 0.9716,
+      "step": 71910
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.456029978521487e-05,
+      "loss": 0.9911,
+      "step": 71920
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.455953813578686e-05,
+      "loss": 1.0864,
+      "step": 71930
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.455877648635886e-05,
+      "loss": 1.0736,
+      "step": 71940
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455801483693086e-05,
+      "loss": 1.0559,
+      "step": 71950
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455725318750286e-05,
+      "loss": 0.9621,
+      "step": 71960
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.4556491538074855e-05,
+      "loss": 1.1448,
+      "step": 71970
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455572988864686e-05,
+      "loss": 1.0979,
+      "step": 71980
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455496823921885e-05,
+      "loss": 1.0858,
+      "step": 71990
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.4554206589790855e-05,
+      "loss": 1.0484,
+      "step": 72000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455344494036285e-05,
+      "loss": 1.0143,
+      "step": 72010
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455268329093485e-05,
+      "loss": 1.0541,
+      "step": 72020
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455192164150685e-05,
+      "loss": 1.0169,
+      "step": 72030
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455115999207885e-05,
+      "loss": 1.0566,
+      "step": 72040
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.455039834265084e-05,
+      "loss": 1.0856,
+      "step": 72050
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.4549636693222846e-05,
+      "loss": 1.0266,
+      "step": 72060
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.4548875043794844e-05,
+      "loss": 1.0874,
+      "step": 72070
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454811339436684e-05,
+      "loss": 1.0116,
+      "step": 72080
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.4547351744938845e-05,
+      "loss": 1.0133,
+      "step": 72090
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454659009551084e-05,
+      "loss": 0.9487,
+      "step": 72100
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454582844608284e-05,
+      "loss": 1.2155,
+      "step": 72110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454506679665484e-05,
+      "loss": 1.0107,
+      "step": 72120
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454430514722684e-05,
+      "loss": 1.0308,
+      "step": 72130
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454354349779883e-05,
+      "loss": 0.9771,
+      "step": 72140
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.454278184837084e-05,
+      "loss": 1.1008,
+      "step": 72150
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.4542020198942834e-05,
+      "loss": 0.9883,
+      "step": 72160
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.454125854951483e-05,
+      "loss": 1.0676,
+      "step": 72170
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.454049690008683e-05,
+      "loss": 1.0171,
+      "step": 72180
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453973525065883e-05,
+      "loss": 1.0923,
+      "step": 72190
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.4538973601230824e-05,
+      "loss": 0.9416,
+      "step": 72200
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453821195180283e-05,
+      "loss": 0.9095,
+      "step": 72210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453745030237482e-05,
+      "loss": 1.0195,
+      "step": 72220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.4536688652946824e-05,
+      "loss": 0.9929,
+      "step": 72230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453592700351882e-05,
+      "loss": 0.9864,
+      "step": 72240
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453516535409082e-05,
+      "loss": 1.027,
+      "step": 72250
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453440370466282e-05,
+      "loss": 1.1893,
+      "step": 72260
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453364205523482e-05,
+      "loss": 0.9972,
+      "step": 72270
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453288040580682e-05,
+      "loss": 1.1574,
+      "step": 72280
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.4532118756378815e-05,
+      "loss": 0.9916,
+      "step": 72290
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453135710695082e-05,
+      "loss": 0.9481,
+      "step": 72300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.453059545752281e-05,
+      "loss": 1.0671,
+      "step": 72310
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.4529833808094814e-05,
+      "loss": 0.9485,
+      "step": 72320
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.452907215866681e-05,
+      "loss": 1.0054,
+      "step": 72330
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.452831050923881e-05,
+      "loss": 1.0719,
+      "step": 72340
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.452754885981081e-05,
+      "loss": 1.0313,
+      "step": 72350
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.452678721038281e-05,
+      "loss": 0.9661,
+      "step": 72360
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.45260255609548e-05,
+      "loss": 0.9961,
+      "step": 72370
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4525263911526806e-05,
+      "loss": 1.0827,
+      "step": 72380
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.45245022620988e-05,
+      "loss": 0.9578,
+      "step": 72390
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.45237406126708e-05,
+      "loss": 1.0159,
+      "step": 72400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.45229789632428e-05,
+      "loss": 1.0138,
+      "step": 72410
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.45222173138148e-05,
+      "loss": 1.1832,
+      "step": 72420
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.452145566438679e-05,
+      "loss": 1.1232,
+      "step": 72430
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.45206940149588e-05,
+      "loss": 1.0133,
+      "step": 72440
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4519932365530795e-05,
+      "loss": 0.9251,
+      "step": 72450
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451917071610279e-05,
+      "loss": 1.0011,
+      "step": 72460
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4518409066674797e-05,
+      "loss": 1.0349,
+      "step": 72470
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4517647417246794e-05,
+      "loss": 1.0643,
+      "step": 72480
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451688576781879e-05,
+      "loss": 1.1432,
+      "step": 72490
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451612411839079e-05,
+      "loss": 1.1074,
+      "step": 72500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451536246896279e-05,
+      "loss": 1.0548,
+      "step": 72510
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4514600819534784e-05,
+      "loss": 0.9794,
+      "step": 72520
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451383917010679e-05,
+      "loss": 0.9581,
+      "step": 72530
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4513077520678786e-05,
+      "loss": 1.1785,
+      "step": 72540
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451231587125078e-05,
+      "loss": 0.9581,
+      "step": 72550
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.451155422182278e-05,
+      "loss": 0.9109,
+      "step": 72560
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4510792572394785e-05,
+      "loss": 1.0993,
+      "step": 72570
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4510030922966776e-05,
+      "loss": 1.1622,
+      "step": 72580
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.450926927353878e-05,
+      "loss": 0.987,
+      "step": 72590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450850762411078e-05,
+      "loss": 0.9549,
+      "step": 72600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4507745974682775e-05,
+      "loss": 1.1355,
+      "step": 72610
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450698432525477e-05,
+      "loss": 1.0674,
+      "step": 72620
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4506222675826776e-05,
+      "loss": 1.1036,
+      "step": 72630
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450546102639877e-05,
+      "loss": 0.9705,
+      "step": 72640
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450469937697077e-05,
+      "loss": 1.0221,
+      "step": 72650
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450393772754277e-05,
+      "loss": 1.0183,
+      "step": 72660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4503176078114766e-05,
+      "loss": 1.0364,
+      "step": 72670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450241442868677e-05,
+      "loss": 0.9659,
+      "step": 72680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450165277925876e-05,
+      "loss": 1.0502,
+      "step": 72690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4500891129830765e-05,
+      "loss": 1.0489,
+      "step": 72700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.450012948040276e-05,
+      "loss": 0.9762,
+      "step": 72710
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449936783097476e-05,
+      "loss": 1.0767,
+      "step": 72720
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449860618154676e-05,
+      "loss": 1.0474,
+      "step": 72730
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449784453211876e-05,
+      "loss": 0.8919,
+      "step": 72740
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449708288269075e-05,
+      "loss": 1.0287,
+      "step": 72750
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449632123326276e-05,
+      "loss": 1.0025,
+      "step": 72760
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4495559583834754e-05,
+      "loss": 0.9736,
+      "step": 72770
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449479793440675e-05,
+      "loss": 1.0151,
+      "step": 72780
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.449403628497875e-05,
+      "loss": 1.0718,
+      "step": 72790
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4493274635550754e-05,
+      "loss": 1.03,
+      "step": 72800
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.4492512986122744e-05,
+      "loss": 1.0491,
+      "step": 72810
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.449175133669475e-05,
+      "loss": 1.0911,
+      "step": 72820
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4490989687266746e-05,
+      "loss": 0.8421,
+      "step": 72830
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4490228037838744e-05,
+      "loss": 1.1749,
+      "step": 72840
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448946638841074e-05,
+      "loss": 1.1568,
+      "step": 72850
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4488704738982745e-05,
+      "loss": 1.0912,
+      "step": 72860
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448794308955474e-05,
+      "loss": 1.1198,
+      "step": 72870
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448718144012674e-05,
+      "loss": 0.9626,
+      "step": 72880
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4486419790698744e-05,
+      "loss": 1.1475,
+      "step": 72890
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4485658141270735e-05,
+      "loss": 1.0098,
+      "step": 72900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448489649184274e-05,
+      "loss": 1.1196,
+      "step": 72910
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448413484241474e-05,
+      "loss": 0.9931,
+      "step": 72920
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4483373192986734e-05,
+      "loss": 1.0951,
+      "step": 72930
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448261154355873e-05,
+      "loss": 1.0492,
+      "step": 72940
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4481849894130736e-05,
+      "loss": 0.9548,
+      "step": 72950
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448108824470273e-05,
+      "loss": 0.9962,
+      "step": 72960
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.448032659527473e-05,
+      "loss": 1.0966,
+      "step": 72970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.447956494584673e-05,
+      "loss": 1.0262,
+      "step": 72980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.4478803296418726e-05,
+      "loss": 1.1336,
+      "step": 72990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.447804164699072e-05,
+      "loss": 1.0263,
+      "step": 73000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.447727999756273e-05,
+      "loss": 1.071,
+      "step": 73010
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.447651834813472e-05,
+      "loss": 1.0564,
+      "step": 73020
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.447575669870672e-05,
+      "loss": 1.0331,
+      "step": 73030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.447499504927872e-05,
+      "loss": 0.9951,
+      "step": 73040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.447423339985072e-05,
+      "loss": 1.1389,
+      "step": 73050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.447347175042272e-05,
+      "loss": 1.0226,
+      "step": 73060
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.447271010099472e-05,
+      "loss": 0.966,
+      "step": 73070
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4471948451566717e-05,
+      "loss": 1.0537,
+      "step": 73080
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4471186802138714e-05,
+      "loss": 0.9873,
+      "step": 73090
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.447042515271072e-05,
+      "loss": 1.1413,
+      "step": 73100
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.446966350328271e-05,
+      "loss": 1.0949,
+      "step": 73110
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.446890185385471e-05,
+      "loss": 1.0417,
+      "step": 73120
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4468140204426704e-05,
+      "loss": 1.1095,
+      "step": 73130
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.446737855499871e-05,
+      "loss": 1.0199,
+      "step": 73140
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4466616905570706e-05,
+      "loss": 0.9865,
+      "step": 73150
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.44658552561427e-05,
+      "loss": 1.091,
+      "step": 73160
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.44650936067147e-05,
+      "loss": 1.0788,
+      "step": 73170
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4464331957286705e-05,
+      "loss": 1.0724,
+      "step": 73180
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4463570307858696e-05,
+      "loss": 1.0721,
+      "step": 73190
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.44628086584307e-05,
+      "loss": 1.0446,
+      "step": 73200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.44620470090027e-05,
+      "loss": 0.9706,
+      "step": 73210
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4461285359574695e-05,
+      "loss": 1.0623,
+      "step": 73220
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.446052371014669e-05,
+      "loss": 1.0686,
+      "step": 73230
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4459762060718696e-05,
+      "loss": 1.0894,
+      "step": 73240
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.4459000411290694e-05,
+      "loss": 0.9851,
+      "step": 73250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445823876186269e-05,
+      "loss": 1.0897,
+      "step": 73260
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4457477112434696e-05,
+      "loss": 0.9958,
+      "step": 73270
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4456715463006686e-05,
+      "loss": 1.151,
+      "step": 73280
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445595381357869e-05,
+      "loss": 1.1027,
+      "step": 73290
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445519216415069e-05,
+      "loss": 0.8995,
+      "step": 73300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4454430514722685e-05,
+      "loss": 1.0026,
+      "step": 73310
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445366886529468e-05,
+      "loss": 1.0182,
+      "step": 73320
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445290721586669e-05,
+      "loss": 0.9603,
+      "step": 73330
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445214556643868e-05,
+      "loss": 1.0291,
+      "step": 73340
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445138391701068e-05,
+      "loss": 1.0333,
+      "step": 73350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.445062226758268e-05,
+      "loss": 1.0288,
+      "step": 73360
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444986061815468e-05,
+      "loss": 0.9795,
+      "step": 73370
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4449098968726674e-05,
+      "loss": 1.0486,
+      "step": 73380
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444833731929868e-05,
+      "loss": 0.9854,
+      "step": 73390
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444757566987067e-05,
+      "loss": 1.0388,
+      "step": 73400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4446814020442674e-05,
+      "loss": 1.1265,
+      "step": 73410
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444605237101467e-05,
+      "loss": 1.1755,
+      "step": 73420
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444529072158667e-05,
+      "loss": 1.0862,
+      "step": 73430
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4444529072158666e-05,
+      "loss": 0.9904,
+      "step": 73440
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444376742273067e-05,
+      "loss": 1.0988,
+      "step": 73450
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.444300577330267e-05,
+      "loss": 1.0263,
+      "step": 73460
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.4442244123874665e-05,
+      "loss": 1.012,
+      "step": 73470
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.444148247444667e-05,
+      "loss": 1.1073,
+      "step": 73480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.444072082501866e-05,
+      "loss": 0.9954,
+      "step": 73490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.4439959175590664e-05,
+      "loss": 1.053,
+      "step": 73500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443919752616266e-05,
+      "loss": 1.0925,
+      "step": 73510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443843587673466e-05,
+      "loss": 0.8669,
+      "step": 73520
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443767422730666e-05,
+      "loss": 1.0626,
+      "step": 73530
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443691257787866e-05,
+      "loss": 1.0124,
+      "step": 73540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443615092845065e-05,
+      "loss": 1.0812,
+      "step": 73550
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.4435389279022656e-05,
+      "loss": 0.9849,
+      "step": 73560
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443462762959465e-05,
+      "loss": 1.1052,
+      "step": 73570
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443386598016665e-05,
+      "loss": 1.1,
+      "step": 73580
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443310433073865e-05,
+      "loss": 0.9243,
+      "step": 73590
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.4432342681310646e-05,
+      "loss": 0.933,
+      "step": 73600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443158103188264e-05,
+      "loss": 1.0516,
+      "step": 73610
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.443081938245465e-05,
+      "loss": 0.9706,
+      "step": 73620
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.4430057733026645e-05,
+      "loss": 1.1285,
+      "step": 73630
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.442929608359864e-05,
+      "loss": 0.9676,
+      "step": 73640
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.442853443417064e-05,
+      "loss": 1.0772,
+      "step": 73650
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.442777278474264e-05,
+      "loss": 0.9673,
+      "step": 73660
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.442701113531464e-05,
+      "loss": 1.1003,
+      "step": 73670
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.442624948588664e-05,
+      "loss": 1.0532,
+      "step": 73680
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.442548783645864e-05,
+      "loss": 1.2058,
+      "step": 73690
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.4424726187030634e-05,
+      "loss": 1.0602,
+      "step": 73700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.442396453760264e-05,
+      "loss": 1.0821,
+      "step": 73710
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.442320288817463e-05,
+      "loss": 1.0551,
+      "step": 73720
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.442244123874663e-05,
+      "loss": 1.03,
+      "step": 73730
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.442167958931863e-05,
+      "loss": 1.0609,
+      "step": 73740
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.442091793989063e-05,
+      "loss": 1.1316,
+      "step": 73750
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.4420156290462626e-05,
+      "loss": 0.9933,
+      "step": 73760
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441939464103463e-05,
+      "loss": 1.0839,
+      "step": 73770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441863299160662e-05,
+      "loss": 0.9956,
+      "step": 73780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.4417871342178625e-05,
+      "loss": 1.116,
+      "step": 73790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441710969275062e-05,
+      "loss": 0.9153,
+      "step": 73800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441634804332262e-05,
+      "loss": 1.0487,
+      "step": 73810
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441558639389462e-05,
+      "loss": 0.9773,
+      "step": 73820
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441482474446662e-05,
+      "loss": 1.0059,
+      "step": 73830
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441406309503862e-05,
+      "loss": 1.0975,
+      "step": 73840
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.4413301445610616e-05,
+      "loss": 1.0741,
+      "step": 73850
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441253979618262e-05,
+      "loss": 1.0428,
+      "step": 73860
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441177814675461e-05,
+      "loss": 0.966,
+      "step": 73870
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.4411016497326616e-05,
+      "loss": 1.0763,
+      "step": 73880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.441025484789861e-05,
+      "loss": 1.1697,
+      "step": 73890
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.440949319847061e-05,
+      "loss": 0.9354,
+      "step": 73900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.440873154904261e-05,
+      "loss": 0.9628,
+      "step": 73910
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.440796989961461e-05,
+      "loss": 1.103,
+      "step": 73920
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.44072082501866e-05,
+      "loss": 1.1103,
+      "step": 73930
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.440644660075861e-05,
+      "loss": 0.9993,
+      "step": 73940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4405684951330605e-05,
+      "loss": 1.0315,
+      "step": 73950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.44049233019026e-05,
+      "loss": 1.0264,
+      "step": 73960
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.44041616524746e-05,
+      "loss": 0.9751,
+      "step": 73970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4403400003046604e-05,
+      "loss": 1.0893,
+      "step": 73980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4402638353618595e-05,
+      "loss": 1.1139,
+      "step": 73990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.44018767041906e-05,
+      "loss": 1.0531,
+      "step": 74000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4401115054762596e-05,
+      "loss": 1.0244,
+      "step": 74010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4400353405334594e-05,
+      "loss": 1.0681,
+      "step": 74020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439959175590659e-05,
+      "loss": 1.03,
+      "step": 74030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439883010647859e-05,
+      "loss": 0.9856,
+      "step": 74040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439806845705059e-05,
+      "loss": 1.0268,
+      "step": 74050
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439730680762259e-05,
+      "loss": 1.0063,
+      "step": 74060
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439654515819459e-05,
+      "loss": 1.0053,
+      "step": 74070
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4395783508766585e-05,
+      "loss": 1.0185,
+      "step": 74080
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439502185933859e-05,
+      "loss": 1.107,
+      "step": 74090
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439426020991058e-05,
+      "loss": 1.1901,
+      "step": 74100
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.4393498560482584e-05,
+      "loss": 0.9561,
+      "step": 74110
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439273691105458e-05,
+      "loss": 0.9846,
+      "step": 74120
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.439197526162658e-05,
+      "loss": 0.9966,
+      "step": 74130
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.439121361219858e-05,
+      "loss": 1.0427,
+      "step": 74140
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.439045196277058e-05,
+      "loss": 1.0395,
+      "step": 74150
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438969031334257e-05,
+      "loss": 1.1253,
+      "step": 74160
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4388928663914576e-05,
+      "loss": 0.9862,
+      "step": 74170
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4388167014486573e-05,
+      "loss": 1.0161,
+      "step": 74180
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438740536505857e-05,
+      "loss": 1.1045,
+      "step": 74190
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438664371563057e-05,
+      "loss": 1.0674,
+      "step": 74200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438588206620257e-05,
+      "loss": 1.1496,
+      "step": 74210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438512041677457e-05,
+      "loss": 1.1367,
+      "step": 74220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438435876734657e-05,
+      "loss": 0.9852,
+      "step": 74230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4383597117918565e-05,
+      "loss": 1.0877,
+      "step": 74240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438283546849056e-05,
+      "loss": 1.0026,
+      "step": 74250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438207381906257e-05,
+      "loss": 1.1294,
+      "step": 74260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4381312169634564e-05,
+      "loss": 1.0196,
+      "step": 74270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.438055052020656e-05,
+      "loss": 1.0853,
+      "step": 74280
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.437978887077856e-05,
+      "loss": 0.9725,
+      "step": 74290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4379027221350563e-05,
+      "loss": 0.9814,
+      "step": 74300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4378265571922554e-05,
+      "loss": 1.0218,
+      "step": 74310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.437750392249456e-05,
+      "loss": 1.1902,
+      "step": 74320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.4376742273066556e-05,
+      "loss": 1.1427,
+      "step": 74330
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.437598062363855e-05,
+      "loss": 1.1383,
+      "step": 74340
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.437521897421055e-05,
+      "loss": 1.095,
+      "step": 74350
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4374457324782555e-05,
+      "loss": 1.0007,
+      "step": 74360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4373695675354546e-05,
+      "loss": 1.0665,
+      "step": 74370
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.437293402592655e-05,
+      "loss": 0.9464,
+      "step": 74380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.437217237649855e-05,
+      "loss": 1.0611,
+      "step": 74390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4371410727070545e-05,
+      "loss": 1.1251,
+      "step": 74400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.437064907764254e-05,
+      "loss": 1.0776,
+      "step": 74410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4369887428214547e-05,
+      "loss": 1.1129,
+      "step": 74420
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4369125778786544e-05,
+      "loss": 0.9549,
+      "step": 74430
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436836412935854e-05,
+      "loss": 0.9989,
+      "step": 74440
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436760247993054e-05,
+      "loss": 1.0514,
+      "step": 74450
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4366840830502536e-05,
+      "loss": 0.9885,
+      "step": 74460
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436607918107454e-05,
+      "loss": 0.941,
+      "step": 74470
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436531753164654e-05,
+      "loss": 1.0812,
+      "step": 74480
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4364555882218536e-05,
+      "loss": 1.0476,
+      "step": 74490
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436379423279053e-05,
+      "loss": 1.0969,
+      "step": 74500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436303258336253e-05,
+      "loss": 1.0995,
+      "step": 74510
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436227093393453e-05,
+      "loss": 1.0043,
+      "step": 74520
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436150928450653e-05,
+      "loss": 1.0391,
+      "step": 74530
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.436074763507852e-05,
+      "loss": 1.1662,
+      "step": 74540
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.435998598565053e-05,
+      "loss": 1.0026,
+      "step": 74550
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.4359224336222525e-05,
+      "loss": 1.0163,
+      "step": 74560
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435846268679452e-05,
+      "loss": 1.054,
+      "step": 74570
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435770103736652e-05,
+      "loss": 1.054,
+      "step": 74580
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4356939387938524e-05,
+      "loss": 0.9294,
+      "step": 74590
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435617773851052e-05,
+      "loss": 1.1162,
+      "step": 74600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435541608908252e-05,
+      "loss": 0.9879,
+      "step": 74610
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4354654439654516e-05,
+      "loss": 0.9272,
+      "step": 74620
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4353892790226514e-05,
+      "loss": 1.0195,
+      "step": 74630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435313114079852e-05,
+      "loss": 1.0125,
+      "step": 74640
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4352369491370515e-05,
+      "loss": 0.9566,
+      "step": 74650
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435160784194251e-05,
+      "loss": 1.041,
+      "step": 74660
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.435084619251451e-05,
+      "loss": 1.1078,
+      "step": 74670
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4350084543086515e-05,
+      "loss": 1.1099,
+      "step": 74680
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4349322893658505e-05,
+      "loss": 1.1555,
+      "step": 74690
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.434856124423051e-05,
+      "loss": 0.9929,
+      "step": 74700
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.434779959480251e-05,
+      "loss": 1.0788,
+      "step": 74710
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4347037945374504e-05,
+      "loss": 1.0363,
+      "step": 74720
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.43462762959465e-05,
+      "loss": 1.1209,
+      "step": 74730
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4345514646518506e-05,
+      "loss": 0.9691,
+      "step": 74740
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.43447529970905e-05,
+      "loss": 0.9275,
+      "step": 74750
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.43439913476625e-05,
+      "loss": 1.1185,
+      "step": 74760
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.43432296982345e-05,
+      "loss": 1.0647,
+      "step": 74770
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.4342468048806496e-05,
+      "loss": 1.0216,
+      "step": 74780
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.4341706399378493e-05,
+      "loss": 1.0716,
+      "step": 74790
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.43409447499505e-05,
+      "loss": 1.0806,
+      "step": 74800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.4340183100522495e-05,
+      "loss": 0.9726,
+      "step": 74810
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433942145109449e-05,
+      "loss": 1.1378,
+      "step": 74820
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433865980166649e-05,
+      "loss": 1.1132,
+      "step": 74830
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433789815223849e-05,
+      "loss": 1.0403,
+      "step": 74840
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433713650281049e-05,
+      "loss": 1.0166,
+      "step": 74850
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433637485338249e-05,
+      "loss": 1.0545,
+      "step": 74860
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433561320395449e-05,
+      "loss": 1.0274,
+      "step": 74870
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.4334851554526484e-05,
+      "loss": 1.0626,
+      "step": 74880
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433408990509849e-05,
+      "loss": 1.058,
+      "step": 74890
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433332825567048e-05,
+      "loss": 1.0895,
+      "step": 74900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.4332566606242483e-05,
+      "loss": 1.1007,
+      "step": 74910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433180495681448e-05,
+      "loss": 0.9738,
+      "step": 74920
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.433104330738648e-05,
+      "loss": 1.0279,
+      "step": 74930
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.4330281657958476e-05,
+      "loss": 0.9406,
+      "step": 74940
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.432952000853048e-05,
+      "loss": 0.9711,
+      "step": 74950
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.432875835910247e-05,
+      "loss": 0.9647,
+      "step": 74960
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.4327996709674475e-05,
+      "loss": 0.9408,
+      "step": 74970
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.432723506024647e-05,
+      "loss": 1.1366,
+      "step": 74980
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.432647341081847e-05,
+      "loss": 0.9883,
+      "step": 74990
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.432571176139047e-05,
+      "loss": 0.9727,
+      "step": 75000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4324950111962465e-05,
+      "loss": 1.103,
+      "step": 75010
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.432418846253447e-05,
+      "loss": 1.0172,
+      "step": 75020
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4323426813106467e-05,
+      "loss": 1.0048,
+      "step": 75030
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4322665163678464e-05,
+      "loss": 1.0842,
+      "step": 75040
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.432190351425046e-05,
+      "loss": 1.0585,
+      "step": 75050
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4321141864822466e-05,
+      "loss": 1.0532,
+      "step": 75060
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4320380215394456e-05,
+      "loss": 1.0202,
+      "step": 75070
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431961856596646e-05,
+      "loss": 1.0173,
+      "step": 75080
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431885691653846e-05,
+      "loss": 1.0018,
+      "step": 75090
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4318095267110456e-05,
+      "loss": 1.0002,
+      "step": 75100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431733361768245e-05,
+      "loss": 0.93,
+      "step": 75110
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431657196825446e-05,
+      "loss": 1.1534,
+      "step": 75120
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431581031882645e-05,
+      "loss": 1.0904,
+      "step": 75130
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431504866939845e-05,
+      "loss": 1.2155,
+      "step": 75140
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431428701997045e-05,
+      "loss": 1.1157,
+      "step": 75150
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431352537054245e-05,
+      "loss": 1.0529,
+      "step": 75160
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4312763721114445e-05,
+      "loss": 1.0004,
+      "step": 75170
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.431200207168645e-05,
+      "loss": 0.9767,
+      "step": 75180
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4311240422258446e-05,
+      "loss": 0.9698,
+      "step": 75190
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.4310478772830444e-05,
+      "loss": 1.12,
+      "step": 75200
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.430971712340244e-05,
+      "loss": 1.0407,
+      "step": 75210
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.430895547397444e-05,
+      "loss": 1.0742,
+      "step": 75220
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430819382454644e-05,
+      "loss": 1.1166,
+      "step": 75230
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430743217511844e-05,
+      "loss": 1.0173,
+      "step": 75240
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430667052569044e-05,
+      "loss": 1.0776,
+      "step": 75250
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.4305908876262435e-05,
+      "loss": 1.0466,
+      "step": 75260
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430514722683444e-05,
+      "loss": 0.9562,
+      "step": 75270
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430438557740643e-05,
+      "loss": 0.984,
+      "step": 75280
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.4303623927978435e-05,
+      "loss": 0.989,
+      "step": 75290
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430286227855043e-05,
+      "loss": 1.032,
+      "step": 75300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430210062912243e-05,
+      "loss": 0.9769,
+      "step": 75310
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430133897969443e-05,
+      "loss": 1.063,
+      "step": 75320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.430057733026643e-05,
+      "loss": 0.962,
+      "step": 75330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429981568083842e-05,
+      "loss": 1.0435,
+      "step": 75340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.4299054031410426e-05,
+      "loss": 1.0589,
+      "step": 75350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.4298292381982424e-05,
+      "loss": 1.0227,
+      "step": 75360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429753073255442e-05,
+      "loss": 1.0556,
+      "step": 75370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429676908312642e-05,
+      "loss": 0.9622,
+      "step": 75380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429600743369842e-05,
+      "loss": 1.0669,
+      "step": 75390
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429524578427042e-05,
+      "loss": 1.0124,
+      "step": 75400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429448413484242e-05,
+      "loss": 1.1285,
+      "step": 75410
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.4293722485414415e-05,
+      "loss": 0.9697,
+      "step": 75420
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429296083598641e-05,
+      "loss": 0.9988,
+      "step": 75430
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.429219918655842e-05,
+      "loss": 1.0543,
+      "step": 75440
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.429143753713041e-05,
+      "loss": 1.023,
+      "step": 75450
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.429067588770241e-05,
+      "loss": 1.1725,
+      "step": 75460
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.428991423827441e-05,
+      "loss": 1.0364,
+      "step": 75470
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.428915258884641e-05,
+      "loss": 1.0044,
+      "step": 75480
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.4288390939418404e-05,
+      "loss": 1.1201,
+      "step": 75490
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.428762928999041e-05,
+      "loss": 1.0418,
+      "step": 75500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.42868676405624e-05,
+      "loss": 1.1449,
+      "step": 75510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.4286105991134403e-05,
+      "loss": 0.9585,
+      "step": 75520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.42853443417064e-05,
+      "loss": 0.9531,
+      "step": 75530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.42845826922784e-05,
+      "loss": 0.9955,
+      "step": 75540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.4283821042850396e-05,
+      "loss": 1.0696,
+      "step": 75550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.42830593934224e-05,
+      "loss": 1.1192,
+      "step": 75560
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.42822977439944e-05,
+      "loss": 1.1066,
+      "step": 75570
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.4281536094566395e-05,
+      "loss": 1.1591,
+      "step": 75580
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.428077444513839e-05,
+      "loss": 1.03,
+      "step": 75590
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.428001279571039e-05,
+      "loss": 1.0536,
+      "step": 75600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.4279251146282394e-05,
+      "loss": 1.1207,
+      "step": 75610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.427848949685439e-05,
+      "loss": 0.9705,
+      "step": 75620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.427772784742639e-05,
+      "loss": 1.0038,
+      "step": 75630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.4276966197998387e-05,
+      "loss": 0.9545,
+      "step": 75640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.427620454857039e-05,
+      "loss": 1.1176,
+      "step": 75650
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.427544289914238e-05,
+      "loss": 1.1544,
+      "step": 75660
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4274681249714386e-05,
+      "loss": 1.0934,
+      "step": 75670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.427391960028638e-05,
+      "loss": 1.0074,
+      "step": 75680
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.427315795085838e-05,
+      "loss": 0.9669,
+      "step": 75690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.427239630143038e-05,
+      "loss": 0.9763,
+      "step": 75700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.427163465200238e-05,
+      "loss": 1.1124,
+      "step": 75710
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.427087300257437e-05,
+      "loss": 0.9397,
+      "step": 75720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.427011135314638e-05,
+      "loss": 1.0461,
+      "step": 75730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4269349703718375e-05,
+      "loss": 1.0415,
+      "step": 75740
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426858805429037e-05,
+      "loss": 0.9397,
+      "step": 75750
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426782640486237e-05,
+      "loss": 1.0044,
+      "step": 75760
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4267064755434374e-05,
+      "loss": 1.0852,
+      "step": 75770
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426630310600637e-05,
+      "loss": 1.1195,
+      "step": 75780
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426554145657837e-05,
+      "loss": 0.9642,
+      "step": 75790
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4264779807150366e-05,
+      "loss": 1.0098,
+      "step": 75800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4264018157722364e-05,
+      "loss": 1.0369,
+      "step": 75810
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426325650829437e-05,
+      "loss": 1.0378,
+      "step": 75820
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4262494858866366e-05,
+      "loss": 0.9224,
+      "step": 75830
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426173320943836e-05,
+      "loss": 1.0369,
+      "step": 75840
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.426097156001036e-05,
+      "loss": 1.0804,
+      "step": 75850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4260209910582365e-05,
+      "loss": 0.9741,
+      "step": 75860
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.4259448261154355e-05,
+      "loss": 1.0242,
+      "step": 75870
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.425868661172636e-05,
+      "loss": 0.9996,
+      "step": 75880
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425792496229835e-05,
+      "loss": 1.0593,
+      "step": 75890
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4257163312870355e-05,
+      "loss": 1.1313,
+      "step": 75900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425640166344235e-05,
+      "loss": 1.1119,
+      "step": 75910
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425564001401435e-05,
+      "loss": 1.0386,
+      "step": 75920
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425487836458635e-05,
+      "loss": 1.0918,
+      "step": 75930
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425411671515835e-05,
+      "loss": 1.0029,
+      "step": 75940
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425335506573034e-05,
+      "loss": 1.0788,
+      "step": 75950
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4252593416302346e-05,
+      "loss": 1.0662,
+      "step": 75960
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4251831766874344e-05,
+      "loss": 0.938,
+      "step": 75970
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.425107011744634e-05,
+      "loss": 0.9562,
+      "step": 75980
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4250308468018345e-05,
+      "loss": 1.0198,
+      "step": 75990
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424954681859034e-05,
+      "loss": 1.0282,
+      "step": 76000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424878516916234e-05,
+      "loss": 1.0443,
+      "step": 76010
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424802351973434e-05,
+      "loss": 1.1866,
+      "step": 76020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424726187030634e-05,
+      "loss": 1.0746,
+      "step": 76030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424650022087833e-05,
+      "loss": 0.9712,
+      "step": 76040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424573857145034e-05,
+      "loss": 1.14,
+      "step": 76050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4244976922022334e-05,
+      "loss": 1.0551,
+      "step": 76060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424421527259433e-05,
+      "loss": 1.0587,
+      "step": 76070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.424345362316633e-05,
+      "loss": 0.9335,
+      "step": 76080
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4242691973738334e-05,
+      "loss": 1.087,
+      "step": 76090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4241930324310324e-05,
+      "loss": 1.0617,
+      "step": 76100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.424116867488233e-05,
+      "loss": 1.0354,
+      "step": 76110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4240407025454326e-05,
+      "loss": 1.1129,
+      "step": 76120
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4239645376026323e-05,
+      "loss": 0.9982,
+      "step": 76130
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423888372659832e-05,
+      "loss": 1.0374,
+      "step": 76140
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4238122077170325e-05,
+      "loss": 1.0412,
+      "step": 76150
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423736042774232e-05,
+      "loss": 0.9878,
+      "step": 76160
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423659877831432e-05,
+      "loss": 1.0843,
+      "step": 76170
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423583712888632e-05,
+      "loss": 1.0753,
+      "step": 76180
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4235075479458315e-05,
+      "loss": 1.0503,
+      "step": 76190
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423431383003032e-05,
+      "loss": 0.9775,
+      "step": 76200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423355218060232e-05,
+      "loss": 1.0508,
+      "step": 76210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4232790531174314e-05,
+      "loss": 1.078,
+      "step": 76220
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.423202888174631e-05,
+      "loss": 0.9929,
+      "step": 76230
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4231267232318316e-05,
+      "loss": 1.0793,
+      "step": 76240
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4230505582890307e-05,
+      "loss": 1.0187,
+      "step": 76250
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.422974393346231e-05,
+      "loss": 1.1104,
+      "step": 76260
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.422898228403431e-05,
+      "loss": 1.1426,
+      "step": 76270
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4228220634606306e-05,
+      "loss": 0.9902,
+      "step": 76280
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.42274589851783e-05,
+      "loss": 0.954,
+      "step": 76290
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.422669733575031e-05,
+      "loss": 1.1432,
+      "step": 76300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.42259356863223e-05,
+      "loss": 1.117,
+      "step": 76310
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.42251740368943e-05,
+      "loss": 0.9277,
+      "step": 76320
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.42244123874663e-05,
+      "loss": 0.9509,
+      "step": 76330
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.42236507380383e-05,
+      "loss": 1.2441,
+      "step": 76340
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4222889088610295e-05,
+      "loss": 1.0215,
+      "step": 76350
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.422212743918229e-05,
+      "loss": 0.936,
+      "step": 76360
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4221365789754297e-05,
+      "loss": 1.0245,
+      "step": 76370
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4220604140326294e-05,
+      "loss": 0.9006,
+      "step": 76380
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421984249089829e-05,
+      "loss": 1.1603,
+      "step": 76390
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421908084147029e-05,
+      "loss": 1.0048,
+      "step": 76400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421831919204229e-05,
+      "loss": 1.0619,
+      "step": 76410
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4217557542614284e-05,
+      "loss": 0.9653,
+      "step": 76420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421679589318629e-05,
+      "loss": 1.0254,
+      "step": 76430
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4216034243758286e-05,
+      "loss": 1.0603,
+      "step": 76440
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421527259433028e-05,
+      "loss": 1.1305,
+      "step": 76450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421451094490228e-05,
+      "loss": 0.9853,
+      "step": 76460
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4213749295474285e-05,
+      "loss": 1.2411,
+      "step": 76470
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4212987646046275e-05,
+      "loss": 1.0056,
+      "step": 76480
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421222599661828e-05,
+      "loss": 1.1174,
+      "step": 76490
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421146434719028e-05,
+      "loss": 0.9833,
+      "step": 76500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4210702697762275e-05,
+      "loss": 1.039,
+      "step": 76510
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.420994104833427e-05,
+      "loss": 0.9817,
+      "step": 76520
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4209179398906276e-05,
+      "loss": 0.9947,
+      "step": 76530
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420841774947827e-05,
+      "loss": 1.0762,
+      "step": 76540
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420765610005027e-05,
+      "loss": 1.027,
+      "step": 76550
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420689445062227e-05,
+      "loss": 1.1197,
+      "step": 76560
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4206132801194266e-05,
+      "loss": 0.9363,
+      "step": 76570
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420537115176627e-05,
+      "loss": 1.0838,
+      "step": 76580
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420460950233827e-05,
+      "loss": 1.0222,
+      "step": 76590
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4203847852910265e-05,
+      "loss": 1.0156,
+      "step": 76600
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420308620348226e-05,
+      "loss": 0.9781,
+      "step": 76610
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420232455405427e-05,
+      "loss": 1.2429,
+      "step": 76620
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420156290462626e-05,
+      "loss": 1.077,
+      "step": 76630
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420080125519826e-05,
+      "loss": 1.0681,
+      "step": 76640
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.420003960577026e-05,
+      "loss": 0.9653,
+      "step": 76650
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419927795634226e-05,
+      "loss": 1.0264,
+      "step": 76660
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4198516306914254e-05,
+      "loss": 0.9851,
+      "step": 76670
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419775465748626e-05,
+      "loss": 1.1885,
+      "step": 76680
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419699300805825e-05,
+      "loss": 0.9548,
+      "step": 76690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4196231358630254e-05,
+      "loss": 1.0682,
+      "step": 76700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419546970920225e-05,
+      "loss": 1.1033,
+      "step": 76710
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419470805977425e-05,
+      "loss": 1.106,
+      "step": 76720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4193946410346246e-05,
+      "loss": 1.0799,
+      "step": 76730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419318476091825e-05,
+      "loss": 0.8943,
+      "step": 76740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.419242311149024e-05,
+      "loss": 1.1089,
+      "step": 76750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4191661462062245e-05,
+      "loss": 1.1592,
+      "step": 76760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.419089981263424e-05,
+      "loss": 1.0149,
+      "step": 76770
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.419013816320624e-05,
+      "loss": 1.1482,
+      "step": 76780
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4189376513778244e-05,
+      "loss": 1.162,
+      "step": 76790
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4188614864350235e-05,
+      "loss": 1.0178,
+      "step": 76800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418785321492224e-05,
+      "loss": 1.0274,
+      "step": 76810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418709156549424e-05,
+      "loss": 1.0386,
+      "step": 76820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4186329916066234e-05,
+      "loss": 0.9866,
+      "step": 76830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418556826663823e-05,
+      "loss": 1.0544,
+      "step": 76840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4184806617210236e-05,
+      "loss": 1.0871,
+      "step": 76850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4184044967782227e-05,
+      "loss": 0.9397,
+      "step": 76860
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418328331835423e-05,
+      "loss": 0.9726,
+      "step": 76870
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418252166892623e-05,
+      "loss": 1.086,
+      "step": 76880
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4181760019498226e-05,
+      "loss": 0.9632,
+      "step": 76890
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418099837007022e-05,
+      "loss": 1.1085,
+      "step": 76900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.418023672064223e-05,
+      "loss": 1.001,
+      "step": 76910
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.417947507121422e-05,
+      "loss": 0.9917,
+      "step": 76920
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.417871342178622e-05,
+      "loss": 1.0696,
+      "step": 76930
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.417795177235822e-05,
+      "loss": 1.0349,
+      "step": 76940
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.417719012293022e-05,
+      "loss": 0.9783,
+      "step": 76950
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.417642847350222e-05,
+      "loss": 1.0696,
+      "step": 76960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.417566682407422e-05,
+      "loss": 1.036,
+      "step": 76970
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4174905174646217e-05,
+      "loss": 1.1415,
+      "step": 76980
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4174143525218214e-05,
+      "loss": 1.0218,
+      "step": 76990
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.417338187579022e-05,
+      "loss": 1.0147,
+      "step": 77000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.417262022636221e-05,
+      "loss": 1.0856,
+      "step": 77010
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.417185857693421e-05,
+      "loss": 1.0459,
+      "step": 77020
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.417109692750621e-05,
+      "loss": 1.0199,
+      "step": 77030
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.417033527807821e-05,
+      "loss": 0.9853,
+      "step": 77040
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4169573628650206e-05,
+      "loss": 1.084,
+      "step": 77050
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.416881197922221e-05,
+      "loss": 1.0356,
+      "step": 77060
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.41680503297942e-05,
+      "loss": 0.8534,
+      "step": 77070
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4167288680366205e-05,
+      "loss": 1.0519,
+      "step": 77080
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.41665270309382e-05,
+      "loss": 0.9669,
+      "step": 77090
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.41657653815102e-05,
+      "loss": 1.0938,
+      "step": 77100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.41650037320822e-05,
+      "loss": 0.8757,
+      "step": 77110
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.41642420826542e-05,
+      "loss": 1.0176,
+      "step": 77120
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.416348043322619e-05,
+      "loss": 1.1769,
+      "step": 77130
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4162718783798196e-05,
+      "loss": 1.0638,
+      "step": 77140
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4161957134370194e-05,
+      "loss": 0.9735,
+      "step": 77150
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.416119548494219e-05,
+      "loss": 1.2217,
+      "step": 77160
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.4160433835514196e-05,
+      "loss": 1.1014,
+      "step": 77170
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.415967218608619e-05,
+      "loss": 1.1706,
+      "step": 77180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.415891053665819e-05,
+      "loss": 1.0045,
+      "step": 77190
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415814888723019e-05,
+      "loss": 1.0533,
+      "step": 77200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415738723780219e-05,
+      "loss": 0.9146,
+      "step": 77210
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415662558837418e-05,
+      "loss": 1.0106,
+      "step": 77220
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415586393894619e-05,
+      "loss": 1.1034,
+      "step": 77230
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.4155102289518185e-05,
+      "loss": 1.0503,
+      "step": 77240
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415434064009018e-05,
+      "loss": 1.0784,
+      "step": 77250
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415357899066218e-05,
+      "loss": 1.1425,
+      "step": 77260
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415281734123418e-05,
+      "loss": 1.0077,
+      "step": 77270
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.4152055691806174e-05,
+      "loss": 1.0615,
+      "step": 77280
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415129404237818e-05,
+      "loss": 1.1463,
+      "step": 77290
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.415053239295017e-05,
+      "loss": 1.02,
+      "step": 77300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.4149770743522174e-05,
+      "loss": 1.023,
+      "step": 77310
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414900909409417e-05,
+      "loss": 1.0996,
+      "step": 77320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414824744466617e-05,
+      "loss": 1.0259,
+      "step": 77330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.4147485795238166e-05,
+      "loss": 1.0119,
+      "step": 77340
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414672414581017e-05,
+      "loss": 0.9861,
+      "step": 77350
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414596249638217e-05,
+      "loss": 1.0354,
+      "step": 77360
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.4145200846954165e-05,
+      "loss": 0.9238,
+      "step": 77370
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414443919752617e-05,
+      "loss": 0.8954,
+      "step": 77380
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414367754809816e-05,
+      "loss": 1.1045,
+      "step": 77390
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.4142915898670164e-05,
+      "loss": 1.0074,
+      "step": 77400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.414215424924216e-05,
+      "loss": 1.0539,
+      "step": 77410
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.414139259981416e-05,
+      "loss": 0.9677,
+      "step": 77420
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.414063095038616e-05,
+      "loss": 1.1129,
+      "step": 77430
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413986930095816e-05,
+      "loss": 1.0776,
+      "step": 77440
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413910765153015e-05,
+      "loss": 0.9326,
+      "step": 77450
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.4138346002102156e-05,
+      "loss": 0.9971,
+      "step": 77460
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413758435267415e-05,
+      "loss": 1.1154,
+      "step": 77470
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413682270324615e-05,
+      "loss": 1.0361,
+      "step": 77480
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413606105381815e-05,
+      "loss": 1.073,
+      "step": 77490
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413529940439015e-05,
+      "loss": 1.0205,
+      "step": 77500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413453775496214e-05,
+      "loss": 1.1355,
+      "step": 77510
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413377610553415e-05,
+      "loss": 1.0495,
+      "step": 77520
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.4133014456106145e-05,
+      "loss": 1.1133,
+      "step": 77530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413225280667814e-05,
+      "loss": 1.1046,
+      "step": 77540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.413149115725014e-05,
+      "loss": 0.9686,
+      "step": 77550
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.4130729507822144e-05,
+      "loss": 1.0139,
+      "step": 77560
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.412996785839414e-05,
+      "loss": 1.0818,
+      "step": 77570
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.412920620896614e-05,
+      "loss": 1.0444,
+      "step": 77580
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.412844455953814e-05,
+      "loss": 1.036,
+      "step": 77590
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.4127682910110134e-05,
+      "loss": 0.9627,
+      "step": 77600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.412692126068214e-05,
+      "loss": 1.0715,
+      "step": 77610
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.4126159611254136e-05,
+      "loss": 1.111,
+      "step": 77620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.412539796182613e-05,
+      "loss": 1.1154,
+      "step": 77630
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.412463631239813e-05,
+      "loss": 1.0424,
+      "step": 77640
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4123874662970135e-05,
+      "loss": 0.97,
+      "step": 77650
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4123113013542126e-05,
+      "loss": 0.9961,
+      "step": 77660
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.412235136411413e-05,
+      "loss": 1.0224,
+      "step": 77670
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.412158971468613e-05,
+      "loss": 1.211,
+      "step": 77680
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4120828065258125e-05,
+      "loss": 1.0731,
+      "step": 77690
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.412006641583012e-05,
+      "loss": 1.0662,
+      "step": 77700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4119304766402126e-05,
+      "loss": 1.0719,
+      "step": 77710
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411854311697412e-05,
+      "loss": 1.032,
+      "step": 77720
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411778146754612e-05,
+      "loss": 1.0053,
+      "step": 77730
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411701981811812e-05,
+      "loss": 1.022,
+      "step": 77740
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4116258168690116e-05,
+      "loss": 1.0764,
+      "step": 77750
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411549651926212e-05,
+      "loss": 1.0777,
+      "step": 77760
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411473486983411e-05,
+      "loss": 1.0662,
+      "step": 77770
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4113973220406116e-05,
+      "loss": 1.0275,
+      "step": 77780
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411321157097811e-05,
+      "loss": 0.9734,
+      "step": 77790
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411244992155011e-05,
+      "loss": 0.9398,
+      "step": 77800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411168827212211e-05,
+      "loss": 1.0334,
+      "step": 77810
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.411092662269411e-05,
+      "loss": 0.9574,
+      "step": 77820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.41101649732661e-05,
+      "loss": 1.1019,
+      "step": 77830
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.410940332383811e-05,
+      "loss": 1.1386,
+      "step": 77840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.4108641674410105e-05,
+      "loss": 1.1364,
+      "step": 77850
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.41078800249821e-05,
+      "loss": 1.0233,
+      "step": 77860
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.41071183755541e-05,
+      "loss": 1.1008,
+      "step": 77870
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4106356726126104e-05,
+      "loss": 1.1785,
+      "step": 77880
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4105595076698094e-05,
+      "loss": 0.9426,
+      "step": 77890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.41048334272701e-05,
+      "loss": 1.1144,
+      "step": 77900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4104071777842096e-05,
+      "loss": 1.0789,
+      "step": 77910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4103310128414094e-05,
+      "loss": 1.1005,
+      "step": 77920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.410254847898609e-05,
+      "loss": 1.1041,
+      "step": 77930
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4101786829558095e-05,
+      "loss": 1.1766,
+      "step": 77940
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.410102518013009e-05,
+      "loss": 1.0268,
+      "step": 77950
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.410026353070209e-05,
+      "loss": 1.0421,
+      "step": 77960
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4099501881274094e-05,
+      "loss": 1.111,
+      "step": 77970
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4098740231846085e-05,
+      "loss": 1.05,
+      "step": 77980
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.409797858241809e-05,
+      "loss": 1.1967,
+      "step": 77990
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.409721693299009e-05,
+      "loss": 1.1105,
+      "step": 78000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4096455283562084e-05,
+      "loss": 1.0892,
+      "step": 78010
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.409569363413408e-05,
+      "loss": 1.1476,
+      "step": 78020
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4094931984706086e-05,
+      "loss": 1.1348,
+      "step": 78030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.409417033527808e-05,
+      "loss": 1.066,
+      "step": 78040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.409340868585008e-05,
+      "loss": 1.1248,
+      "step": 78050
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.409264703642208e-05,
+      "loss": 1.0133,
+      "step": 78060
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.4091885386994076e-05,
+      "loss": 0.9278,
+      "step": 78070
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4091123737566073e-05,
+      "loss": 1.0558,
+      "step": 78080
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.409036208813808e-05,
+      "loss": 0.998,
+      "step": 78090
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408960043871007e-05,
+      "loss": 1.1123,
+      "step": 78100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408883878928207e-05,
+      "loss": 1.0173,
+      "step": 78110
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408807713985407e-05,
+      "loss": 1.0927,
+      "step": 78120
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408731549042607e-05,
+      "loss": 1.038,
+      "step": 78130
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4086553840998065e-05,
+      "loss": 1.0308,
+      "step": 78140
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408579219157007e-05,
+      "loss": 1.0646,
+      "step": 78150
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408503054214207e-05,
+      "loss": 1.0849,
+      "step": 78160
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4084268892714064e-05,
+      "loss": 1.1763,
+      "step": 78170
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408350724328606e-05,
+      "loss": 0.9731,
+      "step": 78180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408274559385806e-05,
+      "loss": 0.9551,
+      "step": 78190
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408198394443006e-05,
+      "loss": 1.0418,
+      "step": 78200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4081222295002054e-05,
+      "loss": 1.1206,
+      "step": 78210
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.408046064557406e-05,
+      "loss": 0.9354,
+      "step": 78220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4079698996146056e-05,
+      "loss": 1.022,
+      "step": 78230
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.407893734671805e-05,
+      "loss": 0.9954,
+      "step": 78240
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.407817569729005e-05,
+      "loss": 1.1302,
+      "step": 78250
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4077414047862055e-05,
+      "loss": 1.122,
+      "step": 78260
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.4076652398434046e-05,
+      "loss": 0.9537,
+      "step": 78270
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.407589074900605e-05,
+      "loss": 1.0238,
+      "step": 78280
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.407512909957805e-05,
+      "loss": 1.0659,
+      "step": 78290
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4074367450150045e-05,
+      "loss": 1.0001,
+      "step": 78300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.407360580072204e-05,
+      "loss": 0.9483,
+      "step": 78310
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4072844151294046e-05,
+      "loss": 0.9555,
+      "step": 78320
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4072082501866044e-05,
+      "loss": 1.1848,
+      "step": 78330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.407132085243804e-05,
+      "loss": 1.0352,
+      "step": 78340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.407055920301004e-05,
+      "loss": 1.0326,
+      "step": 78350
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4069797553582036e-05,
+      "loss": 1.0942,
+      "step": 78360
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406903590415404e-05,
+      "loss": 1.1805,
+      "step": 78370
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406827425472604e-05,
+      "loss": 1.0856,
+      "step": 78380
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4067512605298036e-05,
+      "loss": 1.0885,
+      "step": 78390
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406675095587003e-05,
+      "loss": 1.0248,
+      "step": 78400
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406598930644204e-05,
+      "loss": 1.0332,
+      "step": 78410
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406522765701403e-05,
+      "loss": 1.1166,
+      "step": 78420
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406446600758603e-05,
+      "loss": 1.1243,
+      "step": 78430
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406370435815803e-05,
+      "loss": 0.9113,
+      "step": 78440
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406294270873003e-05,
+      "loss": 1.0149,
+      "step": 78450
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4062181059302025e-05,
+      "loss": 0.9091,
+      "step": 78460
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406141940987403e-05,
+      "loss": 0.9589,
+      "step": 78470
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.406065776044602e-05,
+      "loss": 0.882,
+      "step": 78480
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.4059896111018024e-05,
+      "loss": 1.0779,
+      "step": 78490
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.405913446159002e-05,
+      "loss": 0.9641,
+      "step": 78500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.405837281216202e-05,
+      "loss": 1.0253,
+      "step": 78510
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4057611162734016e-05,
+      "loss": 0.9993,
+      "step": 78520
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405684951330602e-05,
+      "loss": 1.0835,
+      "step": 78530
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405608786387802e-05,
+      "loss": 1.1227,
+      "step": 78540
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4055326214450015e-05,
+      "loss": 1.0675,
+      "step": 78550
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405456456502202e-05,
+      "loss": 1.0723,
+      "step": 78560
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405380291559401e-05,
+      "loss": 0.9307,
+      "step": 78570
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4053041266166014e-05,
+      "loss": 1.0852,
+      "step": 78580
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405227961673801e-05,
+      "loss": 1.0256,
+      "step": 78590
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405151796731001e-05,
+      "loss": 1.1186,
+      "step": 78600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.405075631788201e-05,
+      "loss": 1.0448,
+      "step": 78610
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.404999466845401e-05,
+      "loss": 1.0985,
+      "step": 78620
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4049233019026e-05,
+      "loss": 1.0624,
+      "step": 78630
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4048471369598006e-05,
+      "loss": 1.2002,
+      "step": 78640
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.404770972017e-05,
+      "loss": 0.9862,
+      "step": 78650
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4046948070742e-05,
+      "loss": 1.0962,
+      "step": 78660
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4046186421314e-05,
+      "loss": 1.0239,
+      "step": 78670
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4045424771885996e-05,
+      "loss": 1.0589,
+      "step": 78680
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4044663122457993e-05,
+      "loss": 0.9732,
+      "step": 78690
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.404390147303e-05,
+      "loss": 1.0662,
+      "step": 78700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4043139823601995e-05,
+      "loss": 1.005,
+      "step": 78710
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.404237817417399e-05,
+      "loss": 0.9856,
+      "step": 78720
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.404161652474599e-05,
+      "loss": 0.9309,
+      "step": 78730
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.404085487531799e-05,
+      "loss": 1.064,
+      "step": 78740
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.404009322588999e-05,
+      "loss": 1.1432,
+      "step": 78750
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403933157646199e-05,
+      "loss": 0.9673,
+      "step": 78760
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403856992703399e-05,
+      "loss": 1.0302,
+      "step": 78770
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.4037808277605984e-05,
+      "loss": 0.9076,
+      "step": 78780
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403704662817799e-05,
+      "loss": 1.0296,
+      "step": 78790
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403628497874998e-05,
+      "loss": 0.9559,
+      "step": 78800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403552332932198e-05,
+      "loss": 1.0606,
+      "step": 78810
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403476167989398e-05,
+      "loss": 1.1473,
+      "step": 78820
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403400003046598e-05,
+      "loss": 1.1199,
+      "step": 78830
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.4033238381037976e-05,
+      "loss": 0.9833,
+      "step": 78840
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403247673160998e-05,
+      "loss": 0.9956,
+      "step": 78850
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403171508218197e-05,
+      "loss": 1.1353,
+      "step": 78860
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.4030953432753975e-05,
+      "loss": 1.0429,
+      "step": 78870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.403019178332597e-05,
+      "loss": 0.9974,
+      "step": 78880
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.402943013389797e-05,
+      "loss": 1.0238,
+      "step": 78890
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.402866848446997e-05,
+      "loss": 1.1725,
+      "step": 78900
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.402790683504197e-05,
+      "loss": 0.9749,
+      "step": 78910
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.402714518561397e-05,
+      "loss": 0.9687,
+      "step": 78920
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.4026383536185966e-05,
+      "loss": 1.1052,
+      "step": 78930
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.4025621886757964e-05,
+      "loss": 1.0006,
+      "step": 78940
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.402486023732996e-05,
+      "loss": 1.0267,
+      "step": 78950
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4024098587901966e-05,
+      "loss": 0.9491,
+      "step": 78960
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.402333693847396e-05,
+      "loss": 1.0201,
+      "step": 78970
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.402257528904596e-05,
+      "loss": 1.1461,
+      "step": 78980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.402181363961796e-05,
+      "loss": 0.9735,
+      "step": 78990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.402105199018996e-05,
+      "loss": 0.8763,
+      "step": 79000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.402029034076195e-05,
+      "loss": 1.1142,
+      "step": 79010
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401952869133396e-05,
+      "loss": 1.0319,
+      "step": 79020
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4018767041905955e-05,
+      "loss": 1.107,
+      "step": 79030
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401800539247795e-05,
+      "loss": 0.9993,
+      "step": 79040
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401724374304995e-05,
+      "loss": 1.0367,
+      "step": 79050
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4016482093621954e-05,
+      "loss": 1.0404,
+      "step": 79060
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4015720444193945e-05,
+      "loss": 0.9789,
+      "step": 79070
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401495879476595e-05,
+      "loss": 1.0477,
+      "step": 79080
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4014197145337946e-05,
+      "loss": 0.9171,
+      "step": 79090
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4013435495909944e-05,
+      "loss": 1.0001,
+      "step": 79100
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401267384648194e-05,
+      "loss": 1.0901,
+      "step": 79110
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401191219705394e-05,
+      "loss": 1.0898,
+      "step": 79120
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401115054762594e-05,
+      "loss": 1.0949,
+      "step": 79130
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.401038889819794e-05,
+      "loss": 1.0149,
+      "step": 79140
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.400962724876994e-05,
+      "loss": 1.0046,
+      "step": 79150
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.4008865599341935e-05,
+      "loss": 0.9649,
+      "step": 79160
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400810394991394e-05,
+      "loss": 1.0988,
+      "step": 79170
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400734230048593e-05,
+      "loss": 0.9561,
+      "step": 79180
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.4006580651057935e-05,
+      "loss": 1.0962,
+      "step": 79190
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400581900162993e-05,
+      "loss": 1.0026,
+      "step": 79200
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400505735220193e-05,
+      "loss": 1.0489,
+      "step": 79210
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400429570277393e-05,
+      "loss": 1.0193,
+      "step": 79220
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400353405334593e-05,
+      "loss": 1.0724,
+      "step": 79230
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400277240391792e-05,
+      "loss": 0.9408,
+      "step": 79240
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.4002010754489926e-05,
+      "loss": 0.9818,
+      "step": 79250
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.4001249105061924e-05,
+      "loss": 0.9893,
+      "step": 79260
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.400048745563392e-05,
+      "loss": 1.1274,
+      "step": 79270
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399972580620592e-05,
+      "loss": 1.1208,
+      "step": 79280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399896415677792e-05,
+      "loss": 1.0674,
+      "step": 79290
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399820250734992e-05,
+      "loss": 1.2366,
+      "step": 79300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399744085792192e-05,
+      "loss": 1.0858,
+      "step": 79310
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.3996679208493915e-05,
+      "loss": 0.9284,
+      "step": 79320
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399591755906591e-05,
+      "loss": 1.1326,
+      "step": 79330
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399515590963792e-05,
+      "loss": 1.0662,
+      "step": 79340
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.3994394260209914e-05,
+      "loss": 0.9913,
+      "step": 79350
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399363261078191e-05,
+      "loss": 1.0225,
+      "step": 79360
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.399287096135391e-05,
+      "loss": 1.0904,
+      "step": 79370
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.3992109311925913e-05,
+      "loss": 0.9509,
+      "step": 79380
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3991347662497904e-05,
+      "loss": 1.0012,
+      "step": 79390
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.399058601306991e-05,
+      "loss": 1.0905,
+      "step": 79400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3989824363641906e-05,
+      "loss": 0.941,
+      "step": 79410
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.39890627142139e-05,
+      "loss": 1.1523,
+      "step": 79420
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.39883010647859e-05,
+      "loss": 1.1004,
+      "step": 79430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3987539415357905e-05,
+      "loss": 1.0518,
+      "step": 79440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3986777765929896e-05,
+      "loss": 0.9987,
+      "step": 79450
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.39860161165019e-05,
+      "loss": 1.0476,
+      "step": 79460
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.39852544670739e-05,
+      "loss": 0.9746,
+      "step": 79470
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3984492817645895e-05,
+      "loss": 1.1089,
+      "step": 79480
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.398373116821789e-05,
+      "loss": 0.9624,
+      "step": 79490
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.39829695187899e-05,
+      "loss": 1.0506,
+      "step": 79500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3982207869361894e-05,
+      "loss": 1.0634,
+      "step": 79510
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.398144621993389e-05,
+      "loss": 0.9708,
+      "step": 79520
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.398068457050589e-05,
+      "loss": 0.9556,
+      "step": 79530
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3979922921077887e-05,
+      "loss": 0.9186,
+      "step": 79540
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.397916127164989e-05,
+      "loss": 1.0947,
+      "step": 79550
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.397839962222189e-05,
+      "loss": 1.148,
+      "step": 79560
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3977637972793886e-05,
+      "loss": 1.1278,
+      "step": 79570
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.397687632336588e-05,
+      "loss": 0.9085,
+      "step": 79580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.397611467393788e-05,
+      "loss": 1.0627,
+      "step": 79590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.397535302450988e-05,
+      "loss": 0.9416,
+      "step": 79600
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.397459137508188e-05,
+      "loss": 0.9715,
+      "step": 79610
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.397382972565387e-05,
+      "loss": 1.0605,
+      "step": 79620
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.397306807622588e-05,
+      "loss": 1.0625,
+      "step": 79630
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3972306426797875e-05,
+      "loss": 1.1214,
+      "step": 79640
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.397154477736987e-05,
+      "loss": 0.9719,
+      "step": 79650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.397078312794187e-05,
+      "loss": 1.0207,
+      "step": 79660
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3970021478513874e-05,
+      "loss": 1.0322,
+      "step": 79670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396925982908587e-05,
+      "loss": 1.0006,
+      "step": 79680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396849817965787e-05,
+      "loss": 1.0182,
+      "step": 79690
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3967736530229866e-05,
+      "loss": 1.1062,
+      "step": 79700
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3966974880801864e-05,
+      "loss": 1.078,
+      "step": 79710
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396621323137387e-05,
+      "loss": 1.0335,
+      "step": 79720
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3965451581945865e-05,
+      "loss": 1.0581,
+      "step": 79730
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396468993251786e-05,
+      "loss": 1.1036,
+      "step": 79740
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396392828308986e-05,
+      "loss": 1.1564,
+      "step": 79750
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3963166633661865e-05,
+      "loss": 0.9766,
+      "step": 79760
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3962404984233855e-05,
+      "loss": 1.0727,
+      "step": 79770
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396164333480586e-05,
+      "loss": 1.1405,
+      "step": 79780
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.396088168537786e-05,
+      "loss": 1.0108,
+      "step": 79790
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3960120035949855e-05,
+      "loss": 1.0403,
+      "step": 79800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.395935838652185e-05,
+      "loss": 1.1164,
+      "step": 79810
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3958596737093856e-05,
+      "loss": 1.006,
+      "step": 79820
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395783508766585e-05,
+      "loss": 1.0381,
+      "step": 79830
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395707343823785e-05,
+      "loss": 1.0811,
+      "step": 79840
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395631178880985e-05,
+      "loss": 1.1538,
+      "step": 79850
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.3955550139381846e-05,
+      "loss": 0.9651,
+      "step": 79860
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.3954788489953844e-05,
+      "loss": 0.9951,
+      "step": 79870
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395402684052585e-05,
+      "loss": 1.0478,
+      "step": 79880
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.3953265191097845e-05,
+      "loss": 0.962,
+      "step": 79890
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395250354166984e-05,
+      "loss": 1.0288,
+      "step": 79900
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395174189224184e-05,
+      "loss": 0.9384,
+      "step": 79910
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395098024281384e-05,
+      "loss": 0.9698,
+      "step": 79920
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.395021859338584e-05,
+      "loss": 1.005,
+      "step": 79930
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394945694395784e-05,
+      "loss": 1.0542,
+      "step": 79940
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394869529452984e-05,
+      "loss": 1.1681,
+      "step": 79950
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.3947933645101834e-05,
+      "loss": 0.9012,
+      "step": 79960
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394717199567384e-05,
+      "loss": 1.0488,
+      "step": 79970
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394641034624583e-05,
+      "loss": 0.9909,
+      "step": 79980
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.3945648696817833e-05,
+      "loss": 1.0661,
+      "step": 79990
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394488704738983e-05,
+      "loss": 1.0927,
+      "step": 80000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394412539796183e-05,
+      "loss": 0.9857,
+      "step": 80010
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.3943363748533826e-05,
+      "loss": 0.9577,
+      "step": 80020
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394260209910582e-05,
+      "loss": 0.9644,
+      "step": 80030
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.394184044967782e-05,
+      "loss": 1.111,
+      "step": 80040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3941078800249825e-05,
+      "loss": 1.0734,
+      "step": 80050
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.394031715082182e-05,
+      "loss": 1.1493,
+      "step": 80060
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393955550139382e-05,
+      "loss": 0.9937,
+      "step": 80070
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393879385196582e-05,
+      "loss": 1.0353,
+      "step": 80080
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3938032202537815e-05,
+      "loss": 0.9924,
+      "step": 80090
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393727055310982e-05,
+      "loss": 1.0499,
+      "step": 80100
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393650890368182e-05,
+      "loss": 0.9689,
+      "step": 80110
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3935747254253814e-05,
+      "loss": 0.9927,
+      "step": 80120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393498560482581e-05,
+      "loss": 0.9296,
+      "step": 80130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3934223955397816e-05,
+      "loss": 1.0322,
+      "step": 80140
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3933462305969807e-05,
+      "loss": 1.0833,
+      "step": 80150
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393270065654181e-05,
+      "loss": 1.072,
+      "step": 80160
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.393193900711381e-05,
+      "loss": 1.0023,
+      "step": 80170
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3931177357685806e-05,
+      "loss": 1.0566,
+      "step": 80180
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.39304157082578e-05,
+      "loss": 0.9883,
+      "step": 80190
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.392965405882981e-05,
+      "loss": 1.0516,
+      "step": 80200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.39288924094018e-05,
+      "loss": 1.002,
+      "step": 80210
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.39281307599738e-05,
+      "loss": 1.0337,
+      "step": 80220
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.39273691105458e-05,
+      "loss": 1.0824,
+      "step": 80230
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.39266074611178e-05,
+      "loss": 1.0325,
+      "step": 80240
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.3925845811689795e-05,
+      "loss": 1.0387,
+      "step": 80250
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.39250841622618e-05,
+      "loss": 1.029,
+      "step": 80260
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.3924322512833796e-05,
+      "loss": 1.0526,
+      "step": 80270
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.3923560863405794e-05,
+      "loss": 1.0621,
+      "step": 80280
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.392279921397779e-05,
+      "loss": 1.0592,
+      "step": 80290
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.392203756454979e-05,
+      "loss": 0.9444,
+      "step": 80300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.392127591512179e-05,
+      "loss": 1.0733,
+      "step": 80310
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.392051426569379e-05,
+      "loss": 0.8893,
+      "step": 80320
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391975261626579e-05,
+      "loss": 1.228,
+      "step": 80330
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.3918990966837785e-05,
+      "loss": 0.9285,
+      "step": 80340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391822931740979e-05,
+      "loss": 1.109,
+      "step": 80350
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391746766798178e-05,
+      "loss": 1.0039,
+      "step": 80360
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.3916706018553785e-05,
+      "loss": 1.0497,
+      "step": 80370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391594436912578e-05,
+      "loss": 1.0301,
+      "step": 80380
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391518271969778e-05,
+      "loss": 0.9801,
+      "step": 80390
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391442107026978e-05,
+      "loss": 1.0497,
+      "step": 80400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391365942084178e-05,
+      "loss": 1.0823,
+      "step": 80410
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391289777141377e-05,
+      "loss": 1.0251,
+      "step": 80420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.3912136121985776e-05,
+      "loss": 0.9869,
+      "step": 80430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.3911374472557774e-05,
+      "loss": 1.0295,
+      "step": 80440
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.391061282312977e-05,
+      "loss": 1.3428,
+      "step": 80450
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.390985117370177e-05,
+      "loss": 1.1203,
+      "step": 80460
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.390908952427377e-05,
+      "loss": 1.113,
+      "step": 80470
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.390832787484577e-05,
+      "loss": 1.0478,
+      "step": 80480
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390756622541777e-05,
+      "loss": 0.9839,
+      "step": 80490
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3906804575989765e-05,
+      "loss": 0.8819,
+      "step": 80500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390604292656176e-05,
+      "loss": 1.0915,
+      "step": 80510
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390528127713377e-05,
+      "loss": 1.1964,
+      "step": 80520
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390451962770576e-05,
+      "loss": 1.048,
+      "step": 80530
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390375797827776e-05,
+      "loss": 0.9831,
+      "step": 80540
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390299632884976e-05,
+      "loss": 0.9385,
+      "step": 80550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390223467942176e-05,
+      "loss": 1.0336,
+      "step": 80560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3901473029993754e-05,
+      "loss": 1.0428,
+      "step": 80570
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.390071138056576e-05,
+      "loss": 1.014,
+      "step": 80580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389994973113775e-05,
+      "loss": 1.0965,
+      "step": 80590
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3899188081709754e-05,
+      "loss": 1.0118,
+      "step": 80600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389842643228175e-05,
+      "loss": 1.0428,
+      "step": 80610
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389766478285375e-05,
+      "loss": 1.2839,
+      "step": 80620
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3896903133425746e-05,
+      "loss": 1.0591,
+      "step": 80630
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389614148399775e-05,
+      "loss": 1.0352,
+      "step": 80640
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389537983456975e-05,
+      "loss": 1.0434,
+      "step": 80650
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3894618185141745e-05,
+      "loss": 1.0126,
+      "step": 80660
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389385653571374e-05,
+      "loss": 1.0757,
+      "step": 80670
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389309488628574e-05,
+      "loss": 1.0764,
+      "step": 80680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3892333236857744e-05,
+      "loss": 1.0605,
+      "step": 80690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389157158742974e-05,
+      "loss": 1.0851,
+      "step": 80700
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.389080993800174e-05,
+      "loss": 1.1971,
+      "step": 80710
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.389004828857374e-05,
+      "loss": 1.0501,
+      "step": 80720
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388928663914574e-05,
+      "loss": 1.0416,
+      "step": 80730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388852498971773e-05,
+      "loss": 1.0945,
+      "step": 80740
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3887763340289736e-05,
+      "loss": 1.129,
+      "step": 80750
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388700169086173e-05,
+      "loss": 1.0423,
+      "step": 80760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388624004143373e-05,
+      "loss": 1.0894,
+      "step": 80770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388547839200573e-05,
+      "loss": 0.9042,
+      "step": 80780
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388471674257773e-05,
+      "loss": 1.061,
+      "step": 80790
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388395509314972e-05,
+      "loss": 0.9822,
+      "step": 80800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388319344372173e-05,
+      "loss": 0.9915,
+      "step": 80810
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3882431794293725e-05,
+      "loss": 1.094,
+      "step": 80820
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388167014486572e-05,
+      "loss": 1.0292,
+      "step": 80830
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.388090849543772e-05,
+      "loss": 0.8782,
+      "step": 80840
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3880146846009724e-05,
+      "loss": 1.0299,
+      "step": 80850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.387938519658172e-05,
+      "loss": 0.9644,
+      "step": 80860
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.387862354715372e-05,
+      "loss": 1.0763,
+      "step": 80870
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3877861897725716e-05,
+      "loss": 1.0356,
+      "step": 80880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3877100248297714e-05,
+      "loss": 1.0103,
+      "step": 80890
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.387633859886972e-05,
+      "loss": 1.085,
+      "step": 80900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3875576949441716e-05,
+      "loss": 1.0602,
+      "step": 80910
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.387481530001371e-05,
+      "loss": 1.0631,
+      "step": 80920
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.387405365058571e-05,
+      "loss": 1.171,
+      "step": 80930
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.3873292001157715e-05,
+      "loss": 1.1271,
+      "step": 80940
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.3872530351729705e-05,
+      "loss": 1.0238,
+      "step": 80950
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.387176870230171e-05,
+      "loss": 0.9637,
+      "step": 80960
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.38710070528737e-05,
+      "loss": 1.0975,
+      "step": 80970
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.3870245403445705e-05,
+      "loss": 0.9891,
+      "step": 80980
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.38694837540177e-05,
+      "loss": 0.9758,
+      "step": 80990
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.38687221045897e-05,
+      "loss": 1.1302,
+      "step": 81000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.38679604551617e-05,
+      "loss": 1.1271,
+      "step": 81010
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.38671988057337e-05,
+      "loss": 1.0285,
+      "step": 81020
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.386643715630569e-05,
+      "loss": 1.1282,
+      "step": 81030
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.3865675506877696e-05,
+      "loss": 1.0051,
+      "step": 81040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.3864913857449694e-05,
+      "loss": 1.0136,
+      "step": 81050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.386415220802169e-05,
+      "loss": 0.9768,
+      "step": 81060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.3863390558593695e-05,
+      "loss": 1.0279,
+      "step": 81070
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.386262890916569e-05,
+      "loss": 0.9512,
+      "step": 81080
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.386186725973769e-05,
+      "loss": 0.9849,
+      "step": 81090
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.386110561030969e-05,
+      "loss": 1.0828,
+      "step": 81100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.386034396088169e-05,
+      "loss": 1.1978,
+      "step": 81110
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.385958231145368e-05,
+      "loss": 0.9899,
+      "step": 81120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.385882066202569e-05,
+      "loss": 1.0548,
+      "step": 81130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3858059012597684e-05,
+      "loss": 1.0628,
+      "step": 81140
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.385729736316968e-05,
+      "loss": 1.0165,
+      "step": 81150
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.385653571374168e-05,
+      "loss": 1.0031,
+      "step": 81160
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3855774064313684e-05,
+      "loss": 0.9992,
+      "step": 81170
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3855012414885674e-05,
+      "loss": 1.1241,
+      "step": 81180
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.385425076545768e-05,
+      "loss": 1.0132,
+      "step": 81190
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3853489116029676e-05,
+      "loss": 1.0146,
+      "step": 81200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3852727466601674e-05,
+      "loss": 1.127,
+      "step": 81210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.385196581717367e-05,
+      "loss": 1.0917,
+      "step": 81220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3851204167745675e-05,
+      "loss": 1.0592,
+      "step": 81230
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3850442518317666e-05,
+      "loss": 1.0955,
+      "step": 81240
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384968086888967e-05,
+      "loss": 0.9925,
+      "step": 81250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384891921946167e-05,
+      "loss": 0.9652,
+      "step": 81260
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3848157570033665e-05,
+      "loss": 1.0811,
+      "step": 81270
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384739592060567e-05,
+      "loss": 1.0527,
+      "step": 81280
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384663427117767e-05,
+      "loss": 0.9131,
+      "step": 81290
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3845872621749664e-05,
+      "loss": 0.9779,
+      "step": 81300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384511097232166e-05,
+      "loss": 0.9817,
+      "step": 81310
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.3844349322893666e-05,
+      "loss": 0.9673,
+      "step": 81320
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384358767346566e-05,
+      "loss": 0.9394,
+      "step": 81330
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384282602403766e-05,
+      "loss": 1.2022,
+      "step": 81340
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.384206437460966e-05,
+      "loss": 0.9612,
+      "step": 81350
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3841302725181656e-05,
+      "loss": 1.048,
+      "step": 81360
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.384054107575365e-05,
+      "loss": 0.9019,
+      "step": 81370
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383977942632566e-05,
+      "loss": 1.0015,
+      "step": 81380
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383901777689765e-05,
+      "loss": 1.0987,
+      "step": 81390
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383825612746965e-05,
+      "loss": 1.0959,
+      "step": 81400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383749447804164e-05,
+      "loss": 0.9549,
+      "step": 81410
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383673282861365e-05,
+      "loss": 0.9582,
+      "step": 81420
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3835971179185645e-05,
+      "loss": 1.0143,
+      "step": 81430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383520952975764e-05,
+      "loss": 0.9401,
+      "step": 81440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3834447880329647e-05,
+      "loss": 1.0982,
+      "step": 81450
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3833686230901644e-05,
+      "loss": 1.0925,
+      "step": 81460
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383292458147364e-05,
+      "loss": 1.1657,
+      "step": 81470
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383216293204564e-05,
+      "loss": 1.0491,
+      "step": 81480
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.383140128261764e-05,
+      "loss": 1.1278,
+      "step": 81490
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3830639633189634e-05,
+      "loss": 0.986,
+      "step": 81500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.382987798376164e-05,
+      "loss": 0.9629,
+      "step": 81510
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3829116334333636e-05,
+      "loss": 1.091,
+      "step": 81520
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.382835468490563e-05,
+      "loss": 1.1473,
+      "step": 81530
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.382759303547763e-05,
+      "loss": 1.0522,
+      "step": 81540
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3826831386049635e-05,
+      "loss": 0.9105,
+      "step": 81550
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.3826069736621626e-05,
+      "loss": 0.9703,
+      "step": 81560
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.382530808719363e-05,
+      "loss": 1.0317,
+      "step": 81570
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.382454643776563e-05,
+      "loss": 1.0081,
+      "step": 81580
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.3823784788337625e-05,
+      "loss": 1.1457,
+      "step": 81590
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.382302313890962e-05,
+      "loss": 0.9285,
+      "step": 81600
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.3822261489481626e-05,
+      "loss": 0.9599,
+      "step": 81610
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.382149984005362e-05,
+      "loss": 0.9741,
+      "step": 81620
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.382073819062562e-05,
+      "loss": 1.0968,
+      "step": 81630
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381997654119762e-05,
+      "loss": 1.1189,
+      "step": 81640
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.3819214891769616e-05,
+      "loss": 1.0264,
+      "step": 81650
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381845324234162e-05,
+      "loss": 1.0739,
+      "step": 81660
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381769159291362e-05,
+      "loss": 1.0192,
+      "step": 81670
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.3816929943485615e-05,
+      "loss": 1.076,
+      "step": 81680
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381616829405761e-05,
+      "loss": 0.9591,
+      "step": 81690
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381540664462962e-05,
+      "loss": 1.1166,
+      "step": 81700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381464499520161e-05,
+      "loss": 0.9768,
+      "step": 81710
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381388334577361e-05,
+      "loss": 1.2078,
+      "step": 81720
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381312169634561e-05,
+      "loss": 1.1237,
+      "step": 81730
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381236004691761e-05,
+      "loss": 0.9738,
+      "step": 81740
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.3811598397489604e-05,
+      "loss": 1.0104,
+      "step": 81750
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.381083674806161e-05,
+      "loss": 1.2046,
+      "step": 81760
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.38100750986336e-05,
+      "loss": 1.0762,
+      "step": 81770
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.3809313449205604e-05,
+      "loss": 1.0323,
+      "step": 81780
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.38085517997776e-05,
+      "loss": 1.0812,
+      "step": 81790
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.38077901503496e-05,
+      "loss": 1.1586,
+      "step": 81800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3807028500921596e-05,
+      "loss": 1.0139,
+      "step": 81810
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.38062668514936e-05,
+      "loss": 1.089,
+      "step": 81820
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.380550520206559e-05,
+      "loss": 1.0153,
+      "step": 81830
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3804743552637595e-05,
+      "loss": 1.0382,
+      "step": 81840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.380398190320959e-05,
+      "loss": 1.064,
+      "step": 81850
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.380322025378159e-05,
+      "loss": 1.0428,
+      "step": 81860
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3802458604353594e-05,
+      "loss": 1.0064,
+      "step": 81870
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3801696954925585e-05,
+      "loss": 0.9647,
+      "step": 81880
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.380093530549759e-05,
+      "loss": 1.0732,
+      "step": 81890
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.380017365606959e-05,
+      "loss": 0.9604,
+      "step": 81900
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3799412006641584e-05,
+      "loss": 0.9894,
+      "step": 81910
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379865035721358e-05,
+      "loss": 1.0523,
+      "step": 81920
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3797888707785586e-05,
+      "loss": 1.0205,
+      "step": 81930
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379712705835758e-05,
+      "loss": 1.0378,
+      "step": 81940
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379636540892958e-05,
+      "loss": 0.9267,
+      "step": 81950
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379560375950158e-05,
+      "loss": 1.0503,
+      "step": 81960
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.3794842110073576e-05,
+      "loss": 0.9898,
+      "step": 81970
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379408046064557e-05,
+      "loss": 1.0525,
+      "step": 81980
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379331881121758e-05,
+      "loss": 0.9996,
+      "step": 81990
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379255716178957e-05,
+      "loss": 1.0546,
+      "step": 82000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.379179551236157e-05,
+      "loss": 1.1036,
+      "step": 82010
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.379103386293357e-05,
+      "loss": 1.115,
+      "step": 82020
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.379027221350557e-05,
+      "loss": 1.0881,
+      "step": 82030
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3789510564077565e-05,
+      "loss": 1.0584,
+      "step": 82040
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378874891464957e-05,
+      "loss": 1.0421,
+      "step": 82050
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3787987265221567e-05,
+      "loss": 1.1121,
+      "step": 82060
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3787225615793564e-05,
+      "loss": 0.9904,
+      "step": 82070
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378646396636557e-05,
+      "loss": 0.9169,
+      "step": 82080
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378570231693756e-05,
+      "loss": 0.969,
+      "step": 82090
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378494066750956e-05,
+      "loss": 1.0654,
+      "step": 82100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378417901808156e-05,
+      "loss": 1.1357,
+      "step": 82110
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378341736865356e-05,
+      "loss": 0.9844,
+      "step": 82120
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3782655719225556e-05,
+      "loss": 0.9899,
+      "step": 82130
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378189406979756e-05,
+      "loss": 1.0709,
+      "step": 82140
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.378113242036955e-05,
+      "loss": 1.1131,
+      "step": 82150
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3780370770941555e-05,
+      "loss": 1.0296,
+      "step": 82160
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.377960912151355e-05,
+      "loss": 1.0339,
+      "step": 82170
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.377884747208555e-05,
+      "loss": 1.007,
+      "step": 82180
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.377808582265755e-05,
+      "loss": 1.1409,
+      "step": 82190
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.377732417322955e-05,
+      "loss": 1.0315,
+      "step": 82200
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.377656252380154e-05,
+      "loss": 1.0605,
+      "step": 82210
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3775800874373546e-05,
+      "loss": 0.9092,
+      "step": 82220
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.3775039224945544e-05,
+      "loss": 1.0361,
+      "step": 82230
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377427757551754e-05,
+      "loss": 0.9275,
+      "step": 82240
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3773515926089546e-05,
+      "loss": 1.0014,
+      "step": 82250
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377275427666154e-05,
+      "loss": 1.1663,
+      "step": 82260
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377199262723354e-05,
+      "loss": 1.1091,
+      "step": 82270
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377123097780554e-05,
+      "loss": 1.0754,
+      "step": 82280
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377046932837754e-05,
+      "loss": 0.9553,
+      "step": 82290
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376970767894953e-05,
+      "loss": 0.9585,
+      "step": 82300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376894602952154e-05,
+      "loss": 0.9962,
+      "step": 82310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3768184380093535e-05,
+      "loss": 1.0151,
+      "step": 82320
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376742273066553e-05,
+      "loss": 1.2098,
+      "step": 82330
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376666108123753e-05,
+      "loss": 1.1251,
+      "step": 82340
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376589943180953e-05,
+      "loss": 1.0344,
+      "step": 82350
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3765137782381524e-05,
+      "loss": 1.1842,
+      "step": 82360
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376437613295353e-05,
+      "loss": 1.0067,
+      "step": 82370
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376361448352552e-05,
+      "loss": 0.9877,
+      "step": 82380
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3762852834097524e-05,
+      "loss": 1.0928,
+      "step": 82390
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376209118466952e-05,
+      "loss": 0.9312,
+      "step": 82400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.376132953524152e-05,
+      "loss": 1.1046,
+      "step": 82410
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3760567885813516e-05,
+      "loss": 1.0306,
+      "step": 82420
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.375980623638552e-05,
+      "loss": 1.1035,
+      "step": 82430
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.375904458695752e-05,
+      "loss": 1.0238,
+      "step": 82440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3758282937529515e-05,
+      "loss": 1.0282,
+      "step": 82450
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.375752128810152e-05,
+      "loss": 1.0455,
+      "step": 82460
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.375675963867351e-05,
+      "loss": 1.0517,
+      "step": 82470
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3755997989245514e-05,
+      "loss": 1.0878,
+      "step": 82480
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.375523633981751e-05,
+      "loss": 0.9732,
+      "step": 82490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.375447469038951e-05,
+      "loss": 1.0569,
+      "step": 82500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.375371304096151e-05,
+      "loss": 0.9212,
+      "step": 82510
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.375295139153351e-05,
+      "loss": 1.0176,
+      "step": 82520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.37521897421055e-05,
+      "loss": 0.9715,
+      "step": 82530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3751428092677506e-05,
+      "loss": 0.9375,
+      "step": 82540
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3750666443249503e-05,
+      "loss": 1.0298,
+      "step": 82550
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.37499047938215e-05,
+      "loss": 0.9569,
+      "step": 82560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.37491431443935e-05,
+      "loss": 1.0096,
+      "step": 82570
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.37483814949655e-05,
+      "loss": 0.9412,
+      "step": 82580
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.374761984553749e-05,
+      "loss": 1.1053,
+      "step": 82590
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.37468581961095e-05,
+      "loss": 1.161,
+      "step": 82600
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3746096546681495e-05,
+      "loss": 1.0001,
+      "step": 82610
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.374533489725349e-05,
+      "loss": 0.9633,
+      "step": 82620
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.374457324782549e-05,
+      "loss": 1.0459,
+      "step": 82630
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3743811598397494e-05,
+      "loss": 1.0336,
+      "step": 82640
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.374304994896949e-05,
+      "loss": 0.9562,
+      "step": 82650
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.374228829954149e-05,
+      "loss": 0.9787,
+      "step": 82660
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3741526650113493e-05,
+      "loss": 0.8765,
+      "step": 82670
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3740765000685484e-05,
+      "loss": 1.0585,
+      "step": 82680
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.374000335125749e-05,
+      "loss": 0.9564,
+      "step": 82690
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3739241701829486e-05,
+      "loss": 1.0571,
+      "step": 82700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373848005240148e-05,
+      "loss": 0.9565,
+      "step": 82710
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373771840297348e-05,
+      "loss": 1.1785,
+      "step": 82720
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3736956753545485e-05,
+      "loss": 0.994,
+      "step": 82730
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3736195104117476e-05,
+      "loss": 1.1033,
+      "step": 82740
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373543345468948e-05,
+      "loss": 0.9933,
+      "step": 82750
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373467180526148e-05,
+      "loss": 1.0446,
+      "step": 82760
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3733910155833475e-05,
+      "loss": 1.0151,
+      "step": 82770
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373314850640547e-05,
+      "loss": 0.9958,
+      "step": 82780
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373238685697747e-05,
+      "loss": 0.9899,
+      "step": 82790
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373162520754947e-05,
+      "loss": 1.0503,
+      "step": 82800
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373086355812147e-05,
+      "loss": 1.0087,
+      "step": 82810
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.373010190869347e-05,
+      "loss": 1.0591,
+      "step": 82820
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3729340259265466e-05,
+      "loss": 1.0844,
+      "step": 82830
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3728578609837464e-05,
+      "loss": 1.1053,
+      "step": 82840
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.372781696040946e-05,
+      "loss": 1.0894,
+      "step": 82850
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3727055310981466e-05,
+      "loss": 0.9781,
+      "step": 82860
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.372629366155346e-05,
+      "loss": 1.0455,
+      "step": 82870
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.372553201212546e-05,
+      "loss": 1.0044,
+      "step": 82880
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.372477036269746e-05,
+      "loss": 1.0056,
+      "step": 82890
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.372400871326946e-05,
+      "loss": 1.0486,
+      "step": 82900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.372324706384145e-05,
+      "loss": 0.9446,
+      "step": 82910
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.372248541441346e-05,
+      "loss": 0.9853,
+      "step": 82920
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3721723764985455e-05,
+      "loss": 1.0488,
+      "step": 82930
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.372096211555745e-05,
+      "loss": 1.0044,
+      "step": 82940
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.372020046612945e-05,
+      "loss": 1.0641,
+      "step": 82950
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3719438816701454e-05,
+      "loss": 1.0951,
+      "step": 82960
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3718677167273445e-05,
+      "loss": 1.0625,
+      "step": 82970
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.371791551784545e-05,
+      "loss": 1.0097,
+      "step": 82980
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3717153868417446e-05,
+      "loss": 1.0929,
+      "step": 82990
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3716392218989444e-05,
+      "loss": 1.0349,
+      "step": 83000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.371563056956144e-05,
+      "loss": 0.9828,
+      "step": 83010
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3714868920133445e-05,
+      "loss": 0.9731,
+      "step": 83020
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.371410727070544e-05,
+      "loss": 1.0015,
+      "step": 83030
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.371334562127744e-05,
+      "loss": 1.1361,
+      "step": 83040
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3712583971849445e-05,
+      "loss": 1.0367,
+      "step": 83050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3711822322421435e-05,
+      "loss": 0.9122,
+      "step": 83060
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.371106067299344e-05,
+      "loss": 1.013,
+      "step": 83070
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.371029902356544e-05,
+      "loss": 0.9754,
+      "step": 83080
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.3709537374137434e-05,
+      "loss": 0.9594,
+      "step": 83090
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.370877572470943e-05,
+      "loss": 0.9679,
+      "step": 83100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3708014075281436e-05,
+      "loss": 1.0282,
+      "step": 83110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370725242585343e-05,
+      "loss": 1.0993,
+      "step": 83120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370649077642543e-05,
+      "loss": 0.9251,
+      "step": 83130
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370572912699743e-05,
+      "loss": 1.0567,
+      "step": 83140
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3704967477569426e-05,
+      "loss": 1.0103,
+      "step": 83150
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3704205828141423e-05,
+      "loss": 0.9743,
+      "step": 83160
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370344417871343e-05,
+      "loss": 1.1552,
+      "step": 83170
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370268252928542e-05,
+      "loss": 0.989,
+      "step": 83180
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370192087985742e-05,
+      "loss": 1.0365,
+      "step": 83190
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370115923042942e-05,
+      "loss": 1.0498,
+      "step": 83200
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.370039758100142e-05,
+      "loss": 1.0997,
+      "step": 83210
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3699635931573415e-05,
+      "loss": 1.0734,
+      "step": 83220
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.369887428214542e-05,
+      "loss": 1.0364,
+      "step": 83230
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.369811263271742e-05,
+      "loss": 1.1203,
+      "step": 83240
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3697350983289414e-05,
+      "loss": 0.9674,
+      "step": 83250
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.369658933386141e-05,
+      "loss": 1.0285,
+      "step": 83260
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.369582768443341e-05,
+      "loss": 1.0697,
+      "step": 83270
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3695066035005413e-05,
+      "loss": 1.0747,
+      "step": 83280
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3694304385577404e-05,
+      "loss": 0.9028,
+      "step": 83290
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.369354273614941e-05,
+      "loss": 1.156,
+      "step": 83300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.3692781086721406e-05,
+      "loss": 0.9656,
+      "step": 83310
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.36920194372934e-05,
+      "loss": 1.0397,
+      "step": 83320
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.36912577878654e-05,
+      "loss": 1.0274,
+      "step": 83330
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3690496138437405e-05,
+      "loss": 0.9693,
+      "step": 83340
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3689734489009396e-05,
+      "loss": 0.9862,
+      "step": 83350
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.36889728395814e-05,
+      "loss": 1.0202,
+      "step": 83360
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.36882111901534e-05,
+      "loss": 1.1066,
+      "step": 83370
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3687449540725395e-05,
+      "loss": 0.9901,
+      "step": 83380
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.368668789129739e-05,
+      "loss": 0.9646,
+      "step": 83390
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3685926241869397e-05,
+      "loss": 1.0536,
+      "step": 83400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3685164592441394e-05,
+      "loss": 1.0815,
+      "step": 83410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.368440294301339e-05,
+      "loss": 1.0616,
+      "step": 83420
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.368364129358539e-05,
+      "loss": 1.0869,
+      "step": 83430
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3682879644157386e-05,
+      "loss": 1.1076,
+      "step": 83440
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.368211799472939e-05,
+      "loss": 0.9859,
+      "step": 83450
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.368135634530139e-05,
+      "loss": 1.0972,
+      "step": 83460
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3680594695873386e-05,
+      "loss": 0.9562,
+      "step": 83470
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.367983304644538e-05,
+      "loss": 1.0229,
+      "step": 83480
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.367907139701739e-05,
+      "loss": 1.044,
+      "step": 83490
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.367830974758938e-05,
+      "loss": 1.0183,
+      "step": 83500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.367754809816138e-05,
+      "loss": 1.0257,
+      "step": 83510
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.367678644873338e-05,
+      "loss": 1.0549,
+      "step": 83520
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.367602479930538e-05,
+      "loss": 0.9441,
+      "step": 83530
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.3675263149877375e-05,
+      "loss": 1.1505,
+      "step": 83540
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.367450150044938e-05,
+      "loss": 1.1177,
+      "step": 83550
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.367373985102137e-05,
+      "loss": 1.1514,
+      "step": 83560
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.3672978201593374e-05,
+      "loss": 1.0661,
+      "step": 83570
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.367221655216537e-05,
+      "loss": 1.0143,
+      "step": 83580
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.367145490273737e-05,
+      "loss": 1.143,
+      "step": 83590
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.3670693253309366e-05,
+      "loss": 1.0731,
+      "step": 83600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366993160388137e-05,
+      "loss": 1.0069,
+      "step": 83610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366916995445337e-05,
+      "loss": 0.8563,
+      "step": 83620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.3668408305025365e-05,
+      "loss": 0.8945,
+      "step": 83630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366764665559736e-05,
+      "loss": 0.9299,
+      "step": 83640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366688500616936e-05,
+      "loss": 0.9967,
+      "step": 83650
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.3666123356741365e-05,
+      "loss": 0.9786,
+      "step": 83660
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366536170731336e-05,
+      "loss": 1.1065,
+      "step": 83670
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366460005788536e-05,
+      "loss": 0.9138,
+      "step": 83680
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366383840845736e-05,
+      "loss": 0.9855,
+      "step": 83690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366307675902936e-05,
+      "loss": 0.9277,
+      "step": 83700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366231510960135e-05,
+      "loss": 1.0098,
+      "step": 83710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.3661553460173356e-05,
+      "loss": 0.9425,
+      "step": 83720
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366079181074535e-05,
+      "loss": 0.9509,
+      "step": 83730
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.366003016131735e-05,
+      "loss": 1.0174,
+      "step": 83740
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.365926851188935e-05,
+      "loss": 1.0777,
+      "step": 83750
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.3658506862461346e-05,
+      "loss": 0.9768,
+      "step": 83760
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.3657745213033343e-05,
+      "loss": 1.0407,
+      "step": 83770
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365698356360535e-05,
+      "loss": 1.0604,
+      "step": 83780
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.3656221914177345e-05,
+      "loss": 0.9943,
+      "step": 83790
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365546026474934e-05,
+      "loss": 0.9692,
+      "step": 83800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365469861532134e-05,
+      "loss": 0.993,
+      "step": 83810
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365393696589334e-05,
+      "loss": 1.0796,
+      "step": 83820
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365317531646534e-05,
+      "loss": 0.9666,
+      "step": 83830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365241366703734e-05,
+      "loss": 0.9077,
+      "step": 83840
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365165201760934e-05,
+      "loss": 0.9566,
+      "step": 83850
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.3650890368181334e-05,
+      "loss": 1.1239,
+      "step": 83860
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.365012871875334e-05,
+      "loss": 0.9864,
+      "step": 83870
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364936706932533e-05,
+      "loss": 0.9799,
+      "step": 83880
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.3648605419897333e-05,
+      "loss": 0.9785,
+      "step": 83890
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364784377046933e-05,
+      "loss": 1.0006,
+      "step": 83900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364708212104133e-05,
+      "loss": 1.0546,
+      "step": 83910
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.3646320471613326e-05,
+      "loss": 1.1378,
+      "step": 83920
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364555882218533e-05,
+      "loss": 1.0419,
+      "step": 83930
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364479717275732e-05,
+      "loss": 1.0944,
+      "step": 83940
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.3644035523329325e-05,
+      "loss": 1.1877,
+      "step": 83950
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364327387390132e-05,
+      "loss": 1.2068,
+      "step": 83960
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364251222447332e-05,
+      "loss": 1.1032,
+      "step": 83970
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.364175057504532e-05,
+      "loss": 1.0492,
+      "step": 83980
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.364098892561732e-05,
+      "loss": 1.0387,
+      "step": 83990
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.364022727618932e-05,
+      "loss": 1.0329,
+      "step": 84000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3639465626761317e-05,
+      "loss": 0.9931,
+      "step": 84010
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3638703977333314e-05,
+      "loss": 1.0465,
+      "step": 84020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.363794232790531e-05,
+      "loss": 1.0942,
+      "step": 84030
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3637180678477316e-05,
+      "loss": 1.0488,
+      "step": 84040
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.363641902904931e-05,
+      "loss": 1.1346,
+      "step": 84050
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.363565737962131e-05,
+      "loss": 1.0639,
+      "step": 84060
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.363489573019331e-05,
+      "loss": 1.1485,
+      "step": 84070
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.363413408076531e-05,
+      "loss": 1.0365,
+      "step": 84080
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.36333724313373e-05,
+      "loss": 0.9956,
+      "step": 84090
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.363261078190931e-05,
+      "loss": 0.9846,
+      "step": 84100
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3631849132481305e-05,
+      "loss": 1.0439,
+      "step": 84110
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.36310874830533e-05,
+      "loss": 0.9748,
+      "step": 84120
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.36303258336253e-05,
+      "loss": 1.057,
+      "step": 84130
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3629564184197304e-05,
+      "loss": 1.1265,
+      "step": 84140
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3628802534769295e-05,
+      "loss": 1.0902,
+      "step": 84150
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.36280408853413e-05,
+      "loss": 1.0218,
+      "step": 84160
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3627279235913296e-05,
+      "loss": 1.0395,
+      "step": 84170
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.3626517586485294e-05,
+      "loss": 1.0002,
+      "step": 84180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.362575593705729e-05,
+      "loss": 1.0943,
+      "step": 84190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.362499428762929e-05,
+      "loss": 0.9588,
+      "step": 84200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.362423263820129e-05,
+      "loss": 1.0005,
+      "step": 84210
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.362347098877329e-05,
+      "loss": 1.0135,
+      "step": 84220
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.362270933934529e-05,
+      "loss": 0.9218,
+      "step": 84230
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.3621947689917285e-05,
+      "loss": 1.2128,
+      "step": 84240
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.362118604048929e-05,
+      "loss": 1.0011,
+      "step": 84250
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.362042439106128e-05,
+      "loss": 1.0384,
+      "step": 84260
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.3619662741633285e-05,
+      "loss": 0.9841,
+      "step": 84270
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361890109220528e-05,
+      "loss": 0.9442,
+      "step": 84280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361813944277728e-05,
+      "loss": 0.9333,
+      "step": 84290
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361737779334928e-05,
+      "loss": 1.0074,
+      "step": 84300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361661614392128e-05,
+      "loss": 0.9719,
+      "step": 84310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361585449449327e-05,
+      "loss": 1.1178,
+      "step": 84320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.3615092845065276e-05,
+      "loss": 1.0074,
+      "step": 84330
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.3614331195637274e-05,
+      "loss": 1.0407,
+      "step": 84340
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361356954620927e-05,
+      "loss": 0.9812,
+      "step": 84350
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361280789678127e-05,
+      "loss": 1.0162,
+      "step": 84360
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361204624735327e-05,
+      "loss": 0.9657,
+      "step": 84370
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361128459792527e-05,
+      "loss": 0.8732,
+      "step": 84380
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.361052294849727e-05,
+      "loss": 1.0417,
+      "step": 84390
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.3609761299069265e-05,
+      "loss": 1.0776,
+      "step": 84400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.360899964964126e-05,
+      "loss": 1.0723,
+      "step": 84410
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.360823800021327e-05,
+      "loss": 1.0362,
+      "step": 84420
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3607476350785264e-05,
+      "loss": 1.0878,
+      "step": 84430
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.360671470135726e-05,
+      "loss": 1.0106,
+      "step": 84440
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.360595305192926e-05,
+      "loss": 0.9646,
+      "step": 84450
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3605191402501264e-05,
+      "loss": 1.0389,
+      "step": 84460
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3604429753073254e-05,
+      "loss": 0.9622,
+      "step": 84470
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.360366810364526e-05,
+      "loss": 1.0015,
+      "step": 84480
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3602906454217256e-05,
+      "loss": 1.1303,
+      "step": 84490
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3602144804789253e-05,
+      "loss": 0.9605,
+      "step": 84500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.360138315536125e-05,
+      "loss": 1.0909,
+      "step": 84510
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3600621505933255e-05,
+      "loss": 1.0285,
+      "step": 84520
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3599859856505246e-05,
+      "loss": 0.926,
+      "step": 84530
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359909820707725e-05,
+      "loss": 1.0366,
+      "step": 84540
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359833655764925e-05,
+      "loss": 1.0706,
+      "step": 84550
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3597574908221245e-05,
+      "loss": 0.9842,
+      "step": 84560
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359681325879324e-05,
+      "loss": 0.9732,
+      "step": 84570
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359605160936525e-05,
+      "loss": 1.0472,
+      "step": 84580
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3595289959937244e-05,
+      "loss": 0.97,
+      "step": 84590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359452831050924e-05,
+      "loss": 1.1706,
+      "step": 84600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359376666108124e-05,
+      "loss": 0.9844,
+      "step": 84610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.3593005011653237e-05,
+      "loss": 0.9818,
+      "step": 84620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359224336222524e-05,
+      "loss": 1.0031,
+      "step": 84630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.359148171279723e-05,
+      "loss": 1.0839,
+      "step": 84640
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3590720063369236e-05,
+      "loss": 0.9686,
+      "step": 84650
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358995841394123e-05,
+      "loss": 1.0136,
+      "step": 84660
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358919676451323e-05,
+      "loss": 0.9878,
+      "step": 84670
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358843511508523e-05,
+      "loss": 1.0773,
+      "step": 84680
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358767346565723e-05,
+      "loss": 0.8358,
+      "step": 84690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358691181622922e-05,
+      "loss": 1.0067,
+      "step": 84700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358615016680123e-05,
+      "loss": 1.0001,
+      "step": 84710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3585388517373225e-05,
+      "loss": 1.1058,
+      "step": 84720
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358462686794522e-05,
+      "loss": 1.1252,
+      "step": 84730
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358386521851722e-05,
+      "loss": 1.0261,
+      "step": 84740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3583103569089224e-05,
+      "loss": 1.0552,
+      "step": 84750
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358234191966122e-05,
+      "loss": 0.9617,
+      "step": 84760
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.358158027023322e-05,
+      "loss": 1.0232,
+      "step": 84770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3580818620805216e-05,
+      "loss": 0.9588,
+      "step": 84780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3580056971377214e-05,
+      "loss": 1.0121,
+      "step": 84790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.357929532194922e-05,
+      "loss": 1.0194,
+      "step": 84800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3578533672521216e-05,
+      "loss": 1.1635,
+      "step": 84810
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.357777202309321e-05,
+      "loss": 0.9673,
+      "step": 84820
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.357701037366521e-05,
+      "loss": 1.0875,
+      "step": 84830
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3576248724237215e-05,
+      "loss": 0.9709,
+      "step": 84840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.3575487074809205e-05,
+      "loss": 0.9747,
+      "step": 84850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.357472542538121e-05,
+      "loss": 1.021,
+      "step": 84860
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.357396377595321e-05,
+      "loss": 1.0495,
+      "step": 84870
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3573202126525205e-05,
+      "loss": 0.9428,
+      "step": 84880
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.35724404770972e-05,
+      "loss": 1.0035,
+      "step": 84890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3571678827669206e-05,
+      "loss": 1.0817,
+      "step": 84900
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.35709171782412e-05,
+      "loss": 0.8896,
+      "step": 84910
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.35701555288132e-05,
+      "loss": 1.0222,
+      "step": 84920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.35693938793852e-05,
+      "loss": 0.9915,
+      "step": 84930
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3568632229957196e-05,
+      "loss": 1.0441,
+      "step": 84940
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3567870580529194e-05,
+      "loss": 0.9439,
+      "step": 84950
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.35671089311012e-05,
+      "loss": 1.0102,
+      "step": 84960
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3566347281673195e-05,
+      "loss": 1.0478,
+      "step": 84970
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356558563224519e-05,
+      "loss": 1.0066,
+      "step": 84980
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356482398281719e-05,
+      "loss": 1.0033,
+      "step": 84990
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356406233338919e-05,
+      "loss": 1.0827,
+      "step": 85000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356330068396119e-05,
+      "loss": 1.111,
+      "step": 85010
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356253903453319e-05,
+      "loss": 0.9653,
+      "step": 85020
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356177738510519e-05,
+      "loss": 0.9759,
+      "step": 85030
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3561015735677184e-05,
+      "loss": 1.1322,
+      "step": 85040
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.356025408624919e-05,
+      "loss": 1.0786,
+      "step": 85050
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.355949243682118e-05,
+      "loss": 0.9666,
+      "step": 85060
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.3558730787393184e-05,
+      "loss": 0.9253,
+      "step": 85070
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.355796913796518e-05,
+      "loss": 1.037,
+      "step": 85080
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.355720748853718e-05,
+      "loss": 1.0104,
+      "step": 85090
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3556445839109176e-05,
+      "loss": 1.0723,
+      "step": 85100
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3555684189681173e-05,
+      "loss": 1.0835,
+      "step": 85110
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.355492254025317e-05,
+      "loss": 1.0072,
+      "step": 85120
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3554160890825175e-05,
+      "loss": 0.9292,
+      "step": 85130
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3553399241397166e-05,
+      "loss": 1.0232,
+      "step": 85140
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.355263759196917e-05,
+      "loss": 1.1365,
+      "step": 85150
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.355187594254117e-05,
+      "loss": 0.9364,
+      "step": 85160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3551114293113165e-05,
+      "loss": 1.0763,
+      "step": 85170
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.355035264368517e-05,
+      "loss": 1.0782,
+      "step": 85180
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354959099425717e-05,
+      "loss": 0.9856,
+      "step": 85190
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3548829344829164e-05,
+      "loss": 0.8995,
+      "step": 85200
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354806769540116e-05,
+      "loss": 0.9925,
+      "step": 85210
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3547306045973166e-05,
+      "loss": 1.0838,
+      "step": 85220
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3546544396545157e-05,
+      "loss": 1.0427,
+      "step": 85230
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354578274711716e-05,
+      "loss": 0.9893,
+      "step": 85240
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354502109768916e-05,
+      "loss": 0.8993,
+      "step": 85250
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.3544259448261156e-05,
+      "loss": 1.0158,
+      "step": 85260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354349779883315e-05,
+      "loss": 1.0582,
+      "step": 85270
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354273614940516e-05,
+      "loss": 1.1559,
+      "step": 85280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.354197449997715e-05,
+      "loss": 0.8831,
+      "step": 85290
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.354121285054915e-05,
+      "loss": 0.9516,
+      "step": 85300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.354045120112115e-05,
+      "loss": 1.1144,
+      "step": 85310
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353968955169315e-05,
+      "loss": 1.0602,
+      "step": 85320
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3538927902265145e-05,
+      "loss": 1.0531,
+      "step": 85330
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353816625283715e-05,
+      "loss": 1.0614,
+      "step": 85340
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3537404603409147e-05,
+      "loss": 1.0755,
+      "step": 85350
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3536642953981144e-05,
+      "loss": 1.0035,
+      "step": 85360
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353588130455314e-05,
+      "loss": 1.0438,
+      "step": 85370
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353511965512514e-05,
+      "loss": 1.0994,
+      "step": 85380
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353435800569714e-05,
+      "loss": 0.9226,
+      "step": 85390
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353359635626914e-05,
+      "loss": 1.1137,
+      "step": 85400
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353283470684114e-05,
+      "loss": 1.017,
+      "step": 85410
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3532073057413136e-05,
+      "loss": 0.9313,
+      "step": 85420
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353131140798514e-05,
+      "loss": 0.8571,
+      "step": 85430
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.353054975855713e-05,
+      "loss": 1.1143,
+      "step": 85440
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3529788109129135e-05,
+      "loss": 1.072,
+      "step": 85450
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.352902645970113e-05,
+      "loss": 1.018,
+      "step": 85460
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.352826481027313e-05,
+      "loss": 1.0654,
+      "step": 85470
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.352750316084513e-05,
+      "loss": 1.0334,
+      "step": 85480
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.352674151141713e-05,
+      "loss": 1.1285,
+      "step": 85490
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.352597986198912e-05,
+      "loss": 1.0909,
+      "step": 85500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3525218212561126e-05,
+      "loss": 0.9601,
+      "step": 85510
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3524456563133124e-05,
+      "loss": 1.0603,
+      "step": 85520
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.352369491370512e-05,
+      "loss": 1.0473,
+      "step": 85530
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.352293326427712e-05,
+      "loss": 0.9101,
+      "step": 85540
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.352217161484912e-05,
+      "loss": 1.1153,
+      "step": 85550
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.352140996542112e-05,
+      "loss": 0.867,
+      "step": 85560
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.352064831599312e-05,
+      "loss": 1.0363,
+      "step": 85570
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3519886666565115e-05,
+      "loss": 1.0681,
+      "step": 85580
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351912501713711e-05,
+      "loss": 1.0652,
+      "step": 85590
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351836336770912e-05,
+      "loss": 0.8977,
+      "step": 85600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351760171828111e-05,
+      "loss": 1.1176,
+      "step": 85610
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351684006885311e-05,
+      "loss": 0.9708,
+      "step": 85620
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351607841942511e-05,
+      "loss": 1.1742,
+      "step": 85630
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351531676999711e-05,
+      "loss": 1.0223,
+      "step": 85640
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3514555120569104e-05,
+      "loss": 0.9713,
+      "step": 85650
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.351379347114111e-05,
+      "loss": 0.9263,
+      "step": 85660
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.35130318217131e-05,
+      "loss": 1.0666,
+      "step": 85670
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3512270172285104e-05,
+      "loss": 1.1729,
+      "step": 85680
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.35115085228571e-05,
+      "loss": 0.9637,
+      "step": 85690
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.35107468734291e-05,
+      "loss": 1.1192,
+      "step": 85700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3509985224001096e-05,
+      "loss": 1.1809,
+      "step": 85710
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.35092235745731e-05,
+      "loss": 1.1002,
+      "step": 85720
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.350846192514509e-05,
+      "loss": 0.974,
+      "step": 85730
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3507700275717095e-05,
+      "loss": 1.0801,
+      "step": 85740
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350693862628909e-05,
+      "loss": 1.1048,
+      "step": 85750
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350617697686109e-05,
+      "loss": 0.9675,
+      "step": 85760
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3505415327433094e-05,
+      "loss": 0.9054,
+      "step": 85770
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350465367800509e-05,
+      "loss": 0.9318,
+      "step": 85780
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350389202857709e-05,
+      "loss": 0.9675,
+      "step": 85790
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350313037914909e-05,
+      "loss": 1.3327,
+      "step": 85800
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350236872972109e-05,
+      "loss": 1.0479,
+      "step": 85810
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350160708029308e-05,
+      "loss": 1.0207,
+      "step": 85820
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3500845430865086e-05,
+      "loss": 1.0955,
+      "step": 85830
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.350008378143708e-05,
+      "loss": 0.9756,
+      "step": 85840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349932213200908e-05,
+      "loss": 1.1528,
+      "step": 85850
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349856048258108e-05,
+      "loss": 0.9103,
+      "step": 85860
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349779883315308e-05,
+      "loss": 1.0386,
+      "step": 85870
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349703718372507e-05,
+      "loss": 1.1074,
+      "step": 85880
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349627553429708e-05,
+      "loss": 0.9962,
+      "step": 85890
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3495513884869075e-05,
+      "loss": 0.9798,
+      "step": 85900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349475223544107e-05,
+      "loss": 1.0159,
+      "step": 85910
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349399058601307e-05,
+      "loss": 1.2218,
+      "step": 85920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3493228936585074e-05,
+      "loss": 1.1426,
+      "step": 85930
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3492467287157065e-05,
+      "loss": 1.1077,
+      "step": 85940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.349170563772907e-05,
+      "loss": 0.9778,
+      "step": 85950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3490943988301067e-05,
+      "loss": 1.0764,
+      "step": 85960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3490182338873064e-05,
+      "loss": 0.9059,
+      "step": 85970
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348942068944507e-05,
+      "loss": 1.1004,
+      "step": 85980
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3488659040017066e-05,
+      "loss": 0.9398,
+      "step": 85990
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348789739058906e-05,
+      "loss": 1.0843,
+      "step": 86000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348713574116106e-05,
+      "loss": 1.0976,
+      "step": 86010
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348637409173306e-05,
+      "loss": 1.1358,
+      "step": 86020
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3485612442305056e-05,
+      "loss": 1.0456,
+      "step": 86030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348485079287706e-05,
+      "loss": 1.0373,
+      "step": 86040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348408914344905e-05,
+      "loss": 1.0881,
+      "step": 86050
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3483327494021055e-05,
+      "loss": 0.9951,
+      "step": 86060
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348256584459305e-05,
+      "loss": 0.9929,
+      "step": 86070
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348180419516505e-05,
+      "loss": 1.0447,
+      "step": 86080
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348104254573705e-05,
+      "loss": 1.0093,
+      "step": 86090
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.348028089630905e-05,
+      "loss": 0.9941,
+      "step": 86100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.347951924688104e-05,
+      "loss": 0.9522,
+      "step": 86110
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3478757597453046e-05,
+      "loss": 0.8914,
+      "step": 86120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3477995948025044e-05,
+      "loss": 1.1147,
+      "step": 86130
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.347723429859704e-05,
+      "loss": 0.9244,
+      "step": 86140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3476472649169046e-05,
+      "loss": 1.0674,
+      "step": 86150
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.347571099974104e-05,
+      "loss": 1.0128,
+      "step": 86160
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.347494935031304e-05,
+      "loss": 1.0339,
+      "step": 86170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.347418770088504e-05,
+      "loss": 1.0783,
+      "step": 86180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.347342605145704e-05,
+      "loss": 0.9902,
+      "step": 86190
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.347266440202903e-05,
+      "loss": 0.9442,
+      "step": 86200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.347190275260104e-05,
+      "loss": 1.0521,
+      "step": 86210
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3471141103173035e-05,
+      "loss": 1.0486,
+      "step": 86220
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.347037945374503e-05,
+      "loss": 1.0266,
+      "step": 86230
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.346961780431703e-05,
+      "loss": 0.9408,
+      "step": 86240
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3468856154889034e-05,
+      "loss": 1.2546,
+      "step": 86250
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3468094505461024e-05,
+      "loss": 0.9145,
+      "step": 86260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.346733285603303e-05,
+      "loss": 1.0379,
+      "step": 86270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3466571206605026e-05,
+      "loss": 1.0392,
+      "step": 86280
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3465809557177024e-05,
+      "loss": 0.9324,
+      "step": 86290
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.346504790774902e-05,
+      "loss": 1.0386,
+      "step": 86300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3464286258321025e-05,
+      "loss": 0.955,
+      "step": 86310
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3463524608893016e-05,
+      "loss": 1.1332,
+      "step": 86320
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.346276295946502e-05,
+      "loss": 1.1439,
+      "step": 86330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.346200131003702e-05,
+      "loss": 1.1948,
+      "step": 86340
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3461239660609015e-05,
+      "loss": 0.9276,
+      "step": 86350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.346047801118102e-05,
+      "loss": 0.9823,
+      "step": 86360
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.345971636175302e-05,
+      "loss": 1.1115,
+      "step": 86370
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.3458954712325014e-05,
+      "loss": 0.892,
+      "step": 86380
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.345819306289701e-05,
+      "loss": 1.022,
+      "step": 86390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3457431413469016e-05,
+      "loss": 1.0879,
+      "step": 86400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.345666976404101e-05,
+      "loss": 1.0124,
+      "step": 86410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.345590811461301e-05,
+      "loss": 1.1012,
+      "step": 86420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.345514646518501e-05,
+      "loss": 0.9614,
+      "step": 86430
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3454384815757006e-05,
+      "loss": 0.977,
+      "step": 86440
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3453623166329003e-05,
+      "loss": 1.1122,
+      "step": 86450
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.345286151690101e-05,
+      "loss": 1.021,
+      "step": 86460
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3452099867473e-05,
+      "loss": 1.0965,
+      "step": 86470
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3451338218045e-05,
+      "loss": 0.9963,
+      "step": 86480
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.345057656861699e-05,
+      "loss": 0.9752,
+      "step": 86490
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3449814919189e-05,
+      "loss": 1.0541,
+      "step": 86500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3449053269760995e-05,
+      "loss": 0.9875,
+      "step": 86510
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344829162033299e-05,
+      "loss": 1.1614,
+      "step": 86520
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344752997090499e-05,
+      "loss": 1.0112,
+      "step": 86530
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3446768321476994e-05,
+      "loss": 0.9774,
+      "step": 86540
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344600667204899e-05,
+      "loss": 1.1256,
+      "step": 86550
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344524502262099e-05,
+      "loss": 1.0871,
+      "step": 86560
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344448337319299e-05,
+      "loss": 0.9527,
+      "step": 86570
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3443721723764984e-05,
+      "loss": 1.0894,
+      "step": 86580
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344296007433699e-05,
+      "loss": 0.9881,
+      "step": 86590
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.3442198424908986e-05,
+      "loss": 1.0943,
+      "step": 86600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.344143677548098e-05,
+      "loss": 1.0266,
+      "step": 86610
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.344067512605298e-05,
+      "loss": 1.0295,
+      "step": 86620
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3439913476624985e-05,
+      "loss": 1.1991,
+      "step": 86630
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3439151827196976e-05,
+      "loss": 1.1501,
+      "step": 86640
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343839017776898e-05,
+      "loss": 1.0393,
+      "step": 86650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343762852834098e-05,
+      "loss": 0.9732,
+      "step": 86660
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3436866878912975e-05,
+      "loss": 1.1833,
+      "step": 86670
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343610522948497e-05,
+      "loss": 1.0911,
+      "step": 86680
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3435343580056976e-05,
+      "loss": 0.9842,
+      "step": 86690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343458193062897e-05,
+      "loss": 0.9727,
+      "step": 86700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343382028120097e-05,
+      "loss": 0.9548,
+      "step": 86710
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343305863177297e-05,
+      "loss": 1.1238,
+      "step": 86720
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3432296982344966e-05,
+      "loss": 1.1737,
+      "step": 86730
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3431535332916964e-05,
+      "loss": 1.0211,
+      "step": 86740
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.343077368348897e-05,
+      "loss": 1.067,
+      "step": 86750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3430012034060966e-05,
+      "loss": 1.1373,
+      "step": 86760
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.342925038463296e-05,
+      "loss": 1.0602,
+      "step": 86770
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.342848873520497e-05,
+      "loss": 0.958,
+      "step": 86780
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.342772708577696e-05,
+      "loss": 1.0237,
+      "step": 86790
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.342696543634896e-05,
+      "loss": 1.1195,
+      "step": 86800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.342620378692096e-05,
+      "loss": 1.0953,
+      "step": 86810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.342544213749296e-05,
+      "loss": 0.9922,
+      "step": 86820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3424680488064955e-05,
+      "loss": 0.9815,
+      "step": 86830
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.342391883863696e-05,
+      "loss": 0.9943,
+      "step": 86840
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.342315718920895e-05,
+      "loss": 1.0926,
+      "step": 86850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3422395539780954e-05,
+      "loss": 0.9503,
+      "step": 86860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.342163389035295e-05,
+      "loss": 1.0668,
+      "step": 86870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.342087224092495e-05,
+      "loss": 0.9739,
+      "step": 86880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3420110591496946e-05,
+      "loss": 0.9359,
+      "step": 86890
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341934894206895e-05,
+      "loss": 1.0889,
+      "step": 86900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341858729264094e-05,
+      "loss": 1.1455,
+      "step": 86910
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3417825643212945e-05,
+      "loss": 0.9997,
+      "step": 86920
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341706399378494e-05,
+      "loss": 1.0434,
+      "step": 86930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341630234435694e-05,
+      "loss": 1.1169,
+      "step": 86940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3415540694928944e-05,
+      "loss": 1.0409,
+      "step": 86950
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3414779045500935e-05,
+      "loss": 1.0619,
+      "step": 86960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341401739607294e-05,
+      "loss": 1.0224,
+      "step": 86970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341325574664494e-05,
+      "loss": 1.0306,
+      "step": 86980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3412494097216934e-05,
+      "loss": 1.0433,
+      "step": 86990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341173244778893e-05,
+      "loss": 1.0301,
+      "step": 87000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3410970798360936e-05,
+      "loss": 1.065,
+      "step": 87010
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.341020914893293e-05,
+      "loss": 1.0269,
+      "step": 87020
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.340944749950493e-05,
+      "loss": 1.1179,
+      "step": 87030
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.340868585007693e-05,
+      "loss": 1.0929,
+      "step": 87040
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3407924200648926e-05,
+      "loss": 1.0127,
+      "step": 87050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.3407162551220923e-05,
+      "loss": 1.1948,
+      "step": 87060
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340640090179293e-05,
+      "loss": 1.0327,
+      "step": 87070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340563925236492e-05,
+      "loss": 1.0391,
+      "step": 87080
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340487760293692e-05,
+      "loss": 1.0784,
+      "step": 87090
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340411595350892e-05,
+      "loss": 0.9913,
+      "step": 87100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340335430408092e-05,
+      "loss": 1.1539,
+      "step": 87110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.3402592654652915e-05,
+      "loss": 1.1579,
+      "step": 87120
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340183100522492e-05,
+      "loss": 1.0513,
+      "step": 87130
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.340106935579692e-05,
+      "loss": 1.0957,
+      "step": 87140
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.3400307706368914e-05,
+      "loss": 1.1645,
+      "step": 87150
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.339954605694092e-05,
+      "loss": 1.0176,
+      "step": 87160
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.339878440751291e-05,
+      "loss": 0.9784,
+      "step": 87170
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.339802275808491e-05,
+      "loss": 0.9763,
+      "step": 87180
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.339726110865691e-05,
+      "loss": 1.0106,
+      "step": 87190
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.339649945922891e-05,
+      "loss": 0.9162,
+      "step": 87200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.3395737809800906e-05,
+      "loss": 0.9684,
+      "step": 87210
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.339497616037291e-05,
+      "loss": 1.086,
+      "step": 87220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.33942145109449e-05,
+      "loss": 0.8667,
+      "step": 87230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.3393452861516905e-05,
+      "loss": 0.9505,
+      "step": 87240
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.33926912120889e-05,
+      "loss": 0.9058,
+      "step": 87250
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.33919295626609e-05,
+      "loss": 1.0161,
+      "step": 87260
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.33911679132329e-05,
+      "loss": 1.161,
+      "step": 87270
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.33904062638049e-05,
+      "loss": 1.0768,
+      "step": 87280
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338964461437689e-05,
+      "loss": 1.0672,
+      "step": 87290
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.3388882964948896e-05,
+      "loss": 1.0446,
+      "step": 87300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.3388121315520894e-05,
+      "loss": 1.113,
+      "step": 87310
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338735966609289e-05,
+      "loss": 0.9813,
+      "step": 87320
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338659801666489e-05,
+      "loss": 1.0894,
+      "step": 87330
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338583636723689e-05,
+      "loss": 1.1035,
+      "step": 87340
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338507471780889e-05,
+      "loss": 1.114,
+      "step": 87350
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338431306838089e-05,
+      "loss": 1.0332,
+      "step": 87360
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338355141895289e-05,
+      "loss": 1.091,
+      "step": 87370
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338278976952488e-05,
+      "loss": 1.0948,
+      "step": 87380
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338202812009689e-05,
+      "loss": 1.0595,
+      "step": 87390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338126647066888e-05,
+      "loss": 0.9506,
+      "step": 87400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.338050482124088e-05,
+      "loss": 1.0353,
+      "step": 87410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.337974317181288e-05,
+      "loss": 1.0955,
+      "step": 87420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.337898152238488e-05,
+      "loss": 1.0117,
+      "step": 87430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.3378219872956875e-05,
+      "loss": 1.0856,
+      "step": 87440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.337745822352888e-05,
+      "loss": 0.9167,
+      "step": 87450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.337669657410087e-05,
+      "loss": 0.9626,
+      "step": 87460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.3375934924672874e-05,
+      "loss": 1.0532,
+      "step": 87470
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.337517327524487e-05,
+      "loss": 1.093,
+      "step": 87480
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.337441162581687e-05,
+      "loss": 1.0175,
+      "step": 87490
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3373649976388866e-05,
+      "loss": 1.1377,
+      "step": 87500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.337288832696087e-05,
+      "loss": 1.0803,
+      "step": 87510
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.337212667753287e-05,
+      "loss": 1.0197,
+      "step": 87520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3371365028104865e-05,
+      "loss": 0.9573,
+      "step": 87530
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.337060337867687e-05,
+      "loss": 1.0038,
+      "step": 87540
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336984172924886e-05,
+      "loss": 0.984,
+      "step": 87550
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3369080079820865e-05,
+      "loss": 0.9467,
+      "step": 87560
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336831843039286e-05,
+      "loss": 0.9706,
+      "step": 87570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336755678096486e-05,
+      "loss": 1.0657,
+      "step": 87580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336679513153686e-05,
+      "loss": 1.0165,
+      "step": 87590
+    },
+    {
+      "epoch": 4.0,
+      "eval_cer": 0.624381386692148,
+      "eval_em": 0.007379375591296121,
+      "eval_f1": 0.007379375591296121,
+      "eval_loss": 0.9358349442481995,
+      "eval_runtime": 1008.6933,
+      "eval_samples_per_second": 10.479,
+      "eval_steps_per_second": 1.311,
+      "eval_wer": 0.9926206244087039,
+      "step": 87599
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336603348210886e-05,
+      "loss": 1.0453,
+      "step": 87600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336527183268085e-05,
+      "loss": 1.031,
+      "step": 87610
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3364510183252856e-05,
+      "loss": 1.0414,
+      "step": 87620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3363748533824854e-05,
+      "loss": 1.1265,
+      "step": 87630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336298688439685e-05,
+      "loss": 1.0145,
+      "step": 87640
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336222523496885e-05,
+      "loss": 0.8781,
+      "step": 87650
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.336146358554085e-05,
+      "loss": 1.0509,
+      "step": 87660
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3360701936112843e-05,
+      "loss": 1.0158,
+      "step": 87670
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.335994028668485e-05,
+      "loss": 1.0666,
+      "step": 87680
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.3359178637256845e-05,
+      "loss": 0.9823,
+      "step": 87690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.335841698782884e-05,
+      "loss": 1.0017,
+      "step": 87700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.335765533840084e-05,
+      "loss": 0.9655,
+      "step": 87710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3356893688972844e-05,
+      "loss": 0.9408,
+      "step": 87720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.335613203954484e-05,
+      "loss": 0.9532,
+      "step": 87730
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.335537039011684e-05,
+      "loss": 0.9982,
+      "step": 87740
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3354608740688843e-05,
+      "loss": 1.0269,
+      "step": 87750
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3353847091260834e-05,
+      "loss": 1.0655,
+      "step": 87760
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.335308544183284e-05,
+      "loss": 1.0947,
+      "step": 87770
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3352323792404836e-05,
+      "loss": 0.976,
+      "step": 87780
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.335156214297683e-05,
+      "loss": 1.1241,
+      "step": 87790
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.335080049354883e-05,
+      "loss": 0.9801,
+      "step": 87800
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3350038844120835e-05,
+      "loss": 1.0413,
+      "step": 87810
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3349277194692826e-05,
+      "loss": 1.0285,
+      "step": 87820
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334851554526483e-05,
+      "loss": 1.0376,
+      "step": 87830
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334775389583683e-05,
+      "loss": 0.9818,
+      "step": 87840
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3346992246408825e-05,
+      "loss": 0.9938,
+      "step": 87850
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334623059698082e-05,
+      "loss": 1.0251,
+      "step": 87860
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334546894755282e-05,
+      "loss": 1.1098,
+      "step": 87870
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334470729812482e-05,
+      "loss": 1.0823,
+      "step": 87880
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334394564869682e-05,
+      "loss": 1.0645,
+      "step": 87890
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.334318399926882e-05,
+      "loss": 0.9662,
+      "step": 87900
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3342422349840816e-05,
+      "loss": 1.0169,
+      "step": 87910
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.3341660700412814e-05,
+      "loss": 1.0242,
+      "step": 87920
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.334089905098481e-05,
+      "loss": 0.8936,
+      "step": 87930
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3340137401556816e-05,
+      "loss": 1.049,
+      "step": 87940
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.333937575212881e-05,
+      "loss": 0.968,
+      "step": 87950
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.333861410270081e-05,
+      "loss": 1.0505,
+      "step": 87960
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.333785245327281e-05,
+      "loss": 0.9346,
+      "step": 87970
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.333709080384481e-05,
+      "loss": 1.0398,
+      "step": 87980
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.33363291544168e-05,
+      "loss": 1.0121,
+      "step": 87990
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.333556750498881e-05,
+      "loss": 0.9316,
+      "step": 88000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3334805855560805e-05,
+      "loss": 0.9207,
+      "step": 88010
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.33340442061328e-05,
+      "loss": 0.9615,
+      "step": 88020
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.33332825567048e-05,
+      "loss": 1.0215,
+      "step": 88030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3332520907276804e-05,
+      "loss": 1.0463,
+      "step": 88040
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3331759257848795e-05,
+      "loss": 0.9625,
+      "step": 88050
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.33309976084208e-05,
+      "loss": 0.963,
+      "step": 88060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3330235958992796e-05,
+      "loss": 1.097,
+      "step": 88070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3329474309564794e-05,
+      "loss": 0.9903,
+      "step": 88080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.332871266013679e-05,
+      "loss": 1.1012,
+      "step": 88090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3327951010708795e-05,
+      "loss": 0.9471,
+      "step": 88100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.332718936128079e-05,
+      "loss": 0.8304,
+      "step": 88110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.332642771185279e-05,
+      "loss": 0.9695,
+      "step": 88120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.332566606242479e-05,
+      "loss": 0.9821,
+      "step": 88130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.3324904412996785e-05,
+      "loss": 1.0005,
+      "step": 88140
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.332414276356879e-05,
+      "loss": 0.8913,
+      "step": 88150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.332338111414079e-05,
+      "loss": 0.9522,
+      "step": 88160
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3322619464712785e-05,
+      "loss": 1.0625,
+      "step": 88170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.332185781528478e-05,
+      "loss": 0.9857,
+      "step": 88180
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3321096165856786e-05,
+      "loss": 0.9544,
+      "step": 88190
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.332033451642878e-05,
+      "loss": 0.9426,
+      "step": 88200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331957286700078e-05,
+      "loss": 1.043,
+      "step": 88210
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331881121757278e-05,
+      "loss": 1.0885,
+      "step": 88220
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3318049568144776e-05,
+      "loss": 1.1005,
+      "step": 88230
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3317287918716774e-05,
+      "loss": 1.0803,
+      "step": 88240
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331652626928878e-05,
+      "loss": 1.0237,
+      "step": 88250
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331576461986077e-05,
+      "loss": 0.9625,
+      "step": 88260
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331500297043277e-05,
+      "loss": 1.0773,
+      "step": 88270
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331424132100477e-05,
+      "loss": 0.9052,
+      "step": 88280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331347967157677e-05,
+      "loss": 0.9438,
+      "step": 88290
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3312718022148765e-05,
+      "loss": 0.991,
+      "step": 88300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331195637272077e-05,
+      "loss": 0.9607,
+      "step": 88310
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.331119472329277e-05,
+      "loss": 0.9844,
+      "step": 88320
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3310433073864764e-05,
+      "loss": 0.9422,
+      "step": 88330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.330967142443676e-05,
+      "loss": 0.8777,
+      "step": 88340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.330890977500876e-05,
+      "loss": 1.0046,
+      "step": 88350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.3308148125580763e-05,
+      "loss": 0.9977,
+      "step": 88360
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3307386476152754e-05,
+      "loss": 0.9172,
+      "step": 88370
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.330662482672476e-05,
+      "loss": 1.0122,
+      "step": 88380
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3305863177296756e-05,
+      "loss": 1.0058,
+      "step": 88390
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.330510152786875e-05,
+      "loss": 1.1584,
+      "step": 88400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.330433987844075e-05,
+      "loss": 1.0394,
+      "step": 88410
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3303578229012755e-05,
+      "loss": 1.0744,
+      "step": 88420
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3302816579584746e-05,
+      "loss": 0.9529,
+      "step": 88430
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.330205493015675e-05,
+      "loss": 0.994,
+      "step": 88440
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.330129328072875e-05,
+      "loss": 1.0681,
+      "step": 88450
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3300531631300745e-05,
+      "loss": 1.0492,
+      "step": 88460
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329976998187274e-05,
+      "loss": 0.9521,
+      "step": 88470
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329900833244475e-05,
+      "loss": 1.0742,
+      "step": 88480
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3298246683016744e-05,
+      "loss": 1.1205,
+      "step": 88490
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329748503358874e-05,
+      "loss": 1.0855,
+      "step": 88500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329672338416074e-05,
+      "loss": 1.1096,
+      "step": 88510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3295961734732737e-05,
+      "loss": 0.9293,
+      "step": 88520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329520008530474e-05,
+      "loss": 0.9777,
+      "step": 88530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329443843587674e-05,
+      "loss": 1.0275,
+      "step": 88540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.3293676786448736e-05,
+      "loss": 0.9463,
+      "step": 88550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329291513702073e-05,
+      "loss": 1.1372,
+      "step": 88560
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329215348759274e-05,
+      "loss": 0.9602,
+      "step": 88570
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.329139183816473e-05,
+      "loss": 0.9284,
+      "step": 88580
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.329063018873673e-05,
+      "loss": 0.9698,
+      "step": 88590
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328986853930873e-05,
+      "loss": 1.0609,
+      "step": 88600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328910688988073e-05,
+      "loss": 0.9934,
+      "step": 88610
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.3288345240452725e-05,
+      "loss": 0.9888,
+      "step": 88620
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328758359102473e-05,
+      "loss": 0.9858,
+      "step": 88630
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328682194159672e-05,
+      "loss": 1.1529,
+      "step": 88640
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.3286060292168724e-05,
+      "loss": 0.9783,
+      "step": 88650
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328529864274072e-05,
+      "loss": 1.0722,
+      "step": 88660
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328453699331272e-05,
+      "loss": 0.9436,
+      "step": 88670
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.3283775343884716e-05,
+      "loss": 0.9211,
+      "step": 88680
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328301369445672e-05,
+      "loss": 1.078,
+      "step": 88690
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328225204502872e-05,
+      "loss": 1.0084,
+      "step": 88700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.3281490395600715e-05,
+      "loss": 1.0504,
+      "step": 88710
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.328072874617271e-05,
+      "loss": 1.0504,
+      "step": 88720
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.327996709674471e-05,
+      "loss": 1.0376,
+      "step": 88730
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.3279205447316715e-05,
+      "loss": 1.0697,
+      "step": 88740
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.327844379788871e-05,
+      "loss": 1.1418,
+      "step": 88750
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.327768214846071e-05,
+      "loss": 1.1005,
+      "step": 88760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.327692049903271e-05,
+      "loss": 0.8836,
+      "step": 88770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.327615884960471e-05,
+      "loss": 0.9921,
+      "step": 88780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.32753972001767e-05,
+      "loss": 1.0541,
+      "step": 88790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.3274635550748706e-05,
+      "loss": 1.0737,
+      "step": 88800
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.32738739013207e-05,
+      "loss": 0.8342,
+      "step": 88810
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.32731122518927e-05,
+      "loss": 1.1422,
+      "step": 88820
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.32723506024647e-05,
+      "loss": 0.9144,
+      "step": 88830
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.3271588953036696e-05,
+      "loss": 0.9913,
+      "step": 88840
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.3270827303608694e-05,
+      "loss": 1.0098,
+      "step": 88850
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.32700656541807e-05,
+      "loss": 1.146,
+      "step": 88860
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.3269304004752695e-05,
+      "loss": 1.0103,
+      "step": 88870
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326854235532469e-05,
+      "loss": 0.9592,
+      "step": 88880
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326778070589669e-05,
+      "loss": 1.073,
+      "step": 88890
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326701905646869e-05,
+      "loss": 1.009,
+      "step": 88900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326625740704069e-05,
+      "loss": 1.0374,
+      "step": 88910
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326549575761269e-05,
+      "loss": 0.9616,
+      "step": 88920
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326473410818469e-05,
+      "loss": 1.0542,
+      "step": 88930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.3263972458756684e-05,
+      "loss": 0.9565,
+      "step": 88940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326321080932869e-05,
+      "loss": 0.9535,
+      "step": 88950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326244915990068e-05,
+      "loss": 0.9733,
+      "step": 88960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.3261687510472683e-05,
+      "loss": 0.9767,
+      "step": 88970
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326092586104468e-05,
+      "loss": 0.9588,
+      "step": 88980
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.326016421161668e-05,
+      "loss": 1.0179,
+      "step": 88990
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.3259402562188676e-05,
+      "loss": 0.967,
+      "step": 89000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.325864091276068e-05,
+      "loss": 1.0368,
+      "step": 89010
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.325787926333267e-05,
+      "loss": 0.9643,
+      "step": 89020
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3257117613904675e-05,
+      "loss": 0.9653,
+      "step": 89030
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325635596447667e-05,
+      "loss": 1.1943,
+      "step": 89040
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325559431504867e-05,
+      "loss": 1.064,
+      "step": 89050
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325483266562067e-05,
+      "loss": 1.1429,
+      "step": 89060
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325407101619267e-05,
+      "loss": 1.0282,
+      "step": 89070
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325330936676467e-05,
+      "loss": 1.0325,
+      "step": 89080
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325254771733667e-05,
+      "loss": 0.9343,
+      "step": 89090
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3251786067908664e-05,
+      "loss": 0.9709,
+      "step": 89100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.325102441848066e-05,
+      "loss": 0.9584,
+      "step": 89110
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3250262769052666e-05,
+      "loss": 0.7893,
+      "step": 89120
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324950111962466e-05,
+      "loss": 0.9615,
+      "step": 89130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324873947019666e-05,
+      "loss": 1.0817,
+      "step": 89140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324797782076866e-05,
+      "loss": 0.9459,
+      "step": 89150
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324721617134066e-05,
+      "loss": 1.0377,
+      "step": 89160
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324645452191265e-05,
+      "loss": 0.9219,
+      "step": 89170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324569287248466e-05,
+      "loss": 1.0686,
+      "step": 89180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3244931223056655e-05,
+      "loss": 1.0582,
+      "step": 89190
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324416957362865e-05,
+      "loss": 0.9186,
+      "step": 89200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324340792420065e-05,
+      "loss": 0.9945,
+      "step": 89210
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3242646274772654e-05,
+      "loss": 0.9614,
+      "step": 89220
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3241884625344645e-05,
+      "loss": 0.9625,
+      "step": 89230
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.324112297591665e-05,
+      "loss": 1.1268,
+      "step": 89240
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3240361326488646e-05,
+      "loss": 0.9253,
+      "step": 89250
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3239599677060644e-05,
+      "loss": 1.1549,
+      "step": 89260
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323883802763264e-05,
+      "loss": 1.0206,
+      "step": 89270
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323807637820464e-05,
+      "loss": 1.1693,
+      "step": 89280
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323731472877664e-05,
+      "loss": 1.0974,
+      "step": 89290
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323655307934864e-05,
+      "loss": 1.0439,
+      "step": 89300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323579142992064e-05,
+      "loss": 1.0755,
+      "step": 89310
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3235029780492635e-05,
+      "loss": 1.0561,
+      "step": 89320
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323426813106464e-05,
+      "loss": 0.9248,
+      "step": 89330
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323350648163663e-05,
+      "loss": 1.0537,
+      "step": 89340
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3232744832208635e-05,
+      "loss": 1.057,
+      "step": 89350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323198318278063e-05,
+      "loss": 1.08,
+      "step": 89360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323122153335263e-05,
+      "loss": 1.0326,
+      "step": 89370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.323045988392463e-05,
+      "loss": 1.047,
+      "step": 89380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322969823449663e-05,
+      "loss": 1.0061,
+      "step": 89390
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322893658506862e-05,
+      "loss": 0.9863,
+      "step": 89400
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3228174935640626e-05,
+      "loss": 1.0474,
+      "step": 89410
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3227413286212624e-05,
+      "loss": 1.0404,
+      "step": 89420
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322665163678462e-05,
+      "loss": 0.9609,
+      "step": 89430
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322588998735662e-05,
+      "loss": 1.0881,
+      "step": 89440
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322512833792862e-05,
+      "loss": 0.9984,
+      "step": 89450
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322436668850062e-05,
+      "loss": 1.0141,
+      "step": 89460
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.322360503907262e-05,
+      "loss": 1.0869,
+      "step": 89470
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3222843389644615e-05,
+      "loss": 0.9092,
+      "step": 89480
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.322208174021661e-05,
+      "loss": 1.0774,
+      "step": 89490
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.322132009078862e-05,
+      "loss": 1.074,
+      "step": 89500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3220558441360614e-05,
+      "loss": 1.0085,
+      "step": 89510
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.321979679193261e-05,
+      "loss": 1.0574,
+      "step": 89520
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.321903514250461e-05,
+      "loss": 0.9981,
+      "step": 89530
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3218273493076614e-05,
+      "loss": 1.0716,
+      "step": 89540
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3217511843648604e-05,
+      "loss": 0.9676,
+      "step": 89550
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.321675019422061e-05,
+      "loss": 0.9664,
+      "step": 89560
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3215988544792606e-05,
+      "loss": 0.9188,
+      "step": 89570
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3215226895364604e-05,
+      "loss": 0.9234,
+      "step": 89580
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.32144652459366e-05,
+      "loss": 0.9895,
+      "step": 89590
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3213703596508605e-05,
+      "loss": 0.9889,
+      "step": 89600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3212941947080596e-05,
+      "loss": 1.0155,
+      "step": 89610
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.32121802976526e-05,
+      "loss": 1.0267,
+      "step": 89620
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.32114186482246e-05,
+      "loss": 0.9972,
+      "step": 89630
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3210656998796595e-05,
+      "loss": 0.9996,
+      "step": 89640
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.320989534936859e-05,
+      "loss": 1.1461,
+      "step": 89650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.32091336999406e-05,
+      "loss": 0.9596,
+      "step": 89660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.3208372050512594e-05,
+      "loss": 1.0183,
+      "step": 89670
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320761040108459e-05,
+      "loss": 1.0129,
+      "step": 89680
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320684875165659e-05,
+      "loss": 0.9935,
+      "step": 89690
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320608710222859e-05,
+      "loss": 0.9866,
+      "step": 89700
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320532545280059e-05,
+      "loss": 1.0359,
+      "step": 89710
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320456380337258e-05,
+      "loss": 1.1448,
+      "step": 89720
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.3203802153944586e-05,
+      "loss": 0.9355,
+      "step": 89730
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320304050451658e-05,
+      "loss": 1.0159,
+      "step": 89740
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320227885508858e-05,
+      "loss": 0.9705,
+      "step": 89750
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320151720566058e-05,
+      "loss": 1.0154,
+      "step": 89760
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.320075555623258e-05,
+      "loss": 0.9745,
+      "step": 89770
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319999390680457e-05,
+      "loss": 1.0371,
+      "step": 89780
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319923225737658e-05,
+      "loss": 0.9555,
+      "step": 89790
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.3198470607948575e-05,
+      "loss": 1.0086,
+      "step": 89800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319770895852057e-05,
+      "loss": 1.0374,
+      "step": 89810
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319694730909257e-05,
+      "loss": 0.946,
+      "step": 89820
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.3196185659664574e-05,
+      "loss": 1.0578,
+      "step": 89830
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319542401023657e-05,
+      "loss": 1.1678,
+      "step": 89840
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319466236080857e-05,
+      "loss": 0.9581,
+      "step": 89850
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.3193900711380566e-05,
+      "loss": 1.039,
+      "step": 89860
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.3193139061952564e-05,
+      "loss": 0.9134,
+      "step": 89870
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.319237741252457e-05,
+      "loss": 1.1398,
+      "step": 89880
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.3191615763096566e-05,
+      "loss": 1.0905,
+      "step": 89890
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.319085411366856e-05,
+      "loss": 1.0119,
+      "step": 89900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.319009246424056e-05,
+      "loss": 0.9738,
+      "step": 89910
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3189330814812565e-05,
+      "loss": 1.0103,
+      "step": 89920
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3188569165384556e-05,
+      "loss": 0.8483,
+      "step": 89930
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318780751595656e-05,
+      "loss": 0.9319,
+      "step": 89940
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318704586652856e-05,
+      "loss": 0.9431,
+      "step": 89950
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3186284217100555e-05,
+      "loss": 1.0504,
+      "step": 89960
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318552256767255e-05,
+      "loss": 0.8961,
+      "step": 89970
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3184760918244556e-05,
+      "loss": 0.9635,
+      "step": 89980
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318399926881655e-05,
+      "loss": 0.9691,
+      "step": 89990
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318323761938855e-05,
+      "loss": 0.9343,
+      "step": 90000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318247596996055e-05,
+      "loss": 1.0101,
+      "step": 90010
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3181714320532546e-05,
+      "loss": 0.9196,
+      "step": 90020
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3180952671104544e-05,
+      "loss": 0.8764,
+      "step": 90030
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.318019102167655e-05,
+      "loss": 0.9042,
+      "step": 90040
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.3179429372248545e-05,
+      "loss": 0.9959,
+      "step": 90050
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.317866772282054e-05,
+      "loss": 1.0907,
+      "step": 90060
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.317790607339254e-05,
+      "loss": 0.963,
+      "step": 90070
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.317714442396454e-05,
+      "loss": 0.8825,
+      "step": 90080
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.317638277453654e-05,
+      "loss": 1.1422,
+      "step": 90090
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.317562112510854e-05,
+      "loss": 1.0217,
+      "step": 90100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.317485947568054e-05,
+      "loss": 1.0313,
+      "step": 90110
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3174097826252534e-05,
+      "loss": 0.9839,
+      "step": 90120
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.317333617682454e-05,
+      "loss": 1.0213,
+      "step": 90130
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.317257452739653e-05,
+      "loss": 1.0305,
+      "step": 90140
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3171812877968534e-05,
+      "loss": 0.9607,
+      "step": 90150
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.317105122854053e-05,
+      "loss": 0.9609,
+      "step": 90160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.317028957911253e-05,
+      "loss": 0.9814,
+      "step": 90170
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3169527929684526e-05,
+      "loss": 1.0991,
+      "step": 90180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3168766280256524e-05,
+      "loss": 0.9326,
+      "step": 90190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.316800463082852e-05,
+      "loss": 0.9746,
+      "step": 90200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3167242981400525e-05,
+      "loss": 0.9523,
+      "step": 90210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3166481331972516e-05,
+      "loss": 1.0313,
+      "step": 90220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.316571968254452e-05,
+      "loss": 0.9665,
+      "step": 90230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.316495803311652e-05,
+      "loss": 0.978,
+      "step": 90240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3164196383688515e-05,
+      "loss": 1.0058,
+      "step": 90250
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.316343473426052e-05,
+      "loss": 0.9082,
+      "step": 90260
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.316267308483252e-05,
+      "loss": 0.9707,
+      "step": 90270
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3161911435404514e-05,
+      "loss": 1.0301,
+      "step": 90280
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.316114978597651e-05,
+      "loss": 0.9824,
+      "step": 90290
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3160388136548516e-05,
+      "loss": 0.9743,
+      "step": 90300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.315962648712051e-05,
+      "loss": 0.9588,
+      "step": 90310
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.315886483769251e-05,
+      "loss": 0.9764,
+      "step": 90320
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.315810318826451e-05,
+      "loss": 0.9459,
+      "step": 90330
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.3157341538836506e-05,
+      "loss": 1.0372,
+      "step": 90340
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.31565798894085e-05,
+      "loss": 0.9637,
+      "step": 90350
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.315581823998051e-05,
+      "loss": 1.0697,
+      "step": 90360
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.31550565905525e-05,
+      "loss": 0.9475,
+      "step": 90370
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.31542949411245e-05,
+      "loss": 1.0921,
+      "step": 90380
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.31535332916965e-05,
+      "loss": 0.9166,
+      "step": 90390
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.31527716422685e-05,
+      "loss": 1.0015,
+      "step": 90400
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.3152009992840495e-05,
+      "loss": 1.0229,
+      "step": 90410
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.31512483434125e-05,
+      "loss": 1.0366,
+      "step": 90420
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.315048669398449e-05,
+      "loss": 0.8947,
+      "step": 90430
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.3149725044556494e-05,
+      "loss": 0.8984,
+      "step": 90440
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314896339512849e-05,
+      "loss": 0.9595,
+      "step": 90450
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314820174570049e-05,
+      "loss": 1.0331,
+      "step": 90460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314744009627249e-05,
+      "loss": 0.9888,
+      "step": 90470
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314667844684449e-05,
+      "loss": 0.8943,
+      "step": 90480
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314591679741649e-05,
+      "loss": 0.9125,
+      "step": 90490
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.3145155147988486e-05,
+      "loss": 1.0509,
+      "step": 90500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314439349856049e-05,
+      "loss": 1.1329,
+      "step": 90510
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314363184913248e-05,
+      "loss": 0.9768,
+      "step": 90520
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.3142870199704485e-05,
+      "loss": 1.1197,
+      "step": 90530
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314210855027648e-05,
+      "loss": 1.009,
+      "step": 90540
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.314134690084848e-05,
+      "loss": 0.9566,
+      "step": 90550
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.314058525142048e-05,
+      "loss": 1.0308,
+      "step": 90560
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313982360199248e-05,
+      "loss": 1.0861,
+      "step": 90570
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313906195256447e-05,
+      "loss": 1.085,
+      "step": 90580
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.3138300303136476e-05,
+      "loss": 0.9036,
+      "step": 90590
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.3137538653708474e-05,
+      "loss": 0.8666,
+      "step": 90600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313677700428047e-05,
+      "loss": 0.9365,
+      "step": 90610
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313601535485247e-05,
+      "loss": 1.1064,
+      "step": 90620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.3135253705424466e-05,
+      "loss": 0.9434,
+      "step": 90630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313449205599647e-05,
+      "loss": 0.9542,
+      "step": 90640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313373040656847e-05,
+      "loss": 0.962,
+      "step": 90650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.3132968757140465e-05,
+      "loss": 1.0445,
+      "step": 90660
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313220710771246e-05,
+      "loss": 1.1712,
+      "step": 90670
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313144545828447e-05,
+      "loss": 0.9189,
+      "step": 90680
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.313068380885646e-05,
+      "loss": 1.0481,
+      "step": 90690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.312992215942846e-05,
+      "loss": 0.9323,
+      "step": 90700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.312916051000046e-05,
+      "loss": 0.9319,
+      "step": 90710
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.312839886057246e-05,
+      "loss": 1.0319,
+      "step": 90720
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.3127637211144454e-05,
+      "loss": 1.0627,
+      "step": 90730
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.312687556171646e-05,
+      "loss": 1.0358,
+      "step": 90740
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.312611391228845e-05,
+      "loss": 1.0825,
+      "step": 90750
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.3125352262860454e-05,
+      "loss": 0.8533,
+      "step": 90760
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.312459061343245e-05,
+      "loss": 0.9731,
+      "step": 90770
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.312382896400445e-05,
+      "loss": 1.1253,
+      "step": 90780
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.3123067314576446e-05,
+      "loss": 1.0318,
+      "step": 90790
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.312230566514845e-05,
+      "loss": 0.8702,
+      "step": 90800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.312154401572044e-05,
+      "loss": 1.0354,
+      "step": 90810
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.3120782366292445e-05,
+      "loss": 0.9931,
+      "step": 90820
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.312002071686444e-05,
+      "loss": 1.0315,
+      "step": 90830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311925906743644e-05,
+      "loss": 1.0758,
+      "step": 90840
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.3118497418008444e-05,
+      "loss": 1.1866,
+      "step": 90850
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311773576858044e-05,
+      "loss": 0.9416,
+      "step": 90860
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311697411915244e-05,
+      "loss": 0.9516,
+      "step": 90870
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311621246972444e-05,
+      "loss": 1.0425,
+      "step": 90880
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311545082029644e-05,
+      "loss": 0.9339,
+      "step": 90890
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311468917086843e-05,
+      "loss": 1.0829,
+      "step": 90900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.3113927521440436e-05,
+      "loss": 0.939,
+      "step": 90910
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.3113165872012433e-05,
+      "loss": 0.9121,
+      "step": 90920
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311240422258443e-05,
+      "loss": 0.9903,
+      "step": 90930
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311164257315643e-05,
+      "loss": 0.9789,
+      "step": 90940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311088092372843e-05,
+      "loss": 1.04,
+      "step": 90950
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.311011927430042e-05,
+      "loss": 1.0249,
+      "step": 90960
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.310935762487243e-05,
+      "loss": 0.9215,
+      "step": 90970
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.3108595975444425e-05,
+      "loss": 0.9571,
+      "step": 90980
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.310783432601642e-05,
+      "loss": 0.9721,
+      "step": 90990
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.310707267658842e-05,
+      "loss": 0.9361,
+      "step": 91000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3106311027160424e-05,
+      "loss": 0.9647,
+      "step": 91010
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3105549377732415e-05,
+      "loss": 0.9526,
+      "step": 91020
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.310478772830442e-05,
+      "loss": 0.9295,
+      "step": 91030
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.310402607887642e-05,
+      "loss": 1.0494,
+      "step": 91040
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3103264429448414e-05,
+      "loss": 1.0247,
+      "step": 91050
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.310250278002042e-05,
+      "loss": 0.9565,
+      "step": 91060
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3101741130592416e-05,
+      "loss": 0.9581,
+      "step": 91070
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.310097948116441e-05,
+      "loss": 0.9237,
+      "step": 91080
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.310021783173641e-05,
+      "loss": 0.9977,
+      "step": 91090
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.309945618230841e-05,
+      "loss": 1.0455,
+      "step": 91100
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3098694532880406e-05,
+      "loss": 0.9255,
+      "step": 91110
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.309793288345241e-05,
+      "loss": 1.1214,
+      "step": 91120
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.30971712340244e-05,
+      "loss": 0.945,
+      "step": 91130
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3096409584596405e-05,
+      "loss": 1.0643,
+      "step": 91140
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.30956479351684e-05,
+      "loss": 0.9981,
+      "step": 91150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.30948862857404e-05,
+      "loss": 1.0795,
+      "step": 91160
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.30941246363124e-05,
+      "loss": 0.9063,
+      "step": 91170
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.30933629868844e-05,
+      "loss": 1.0122,
+      "step": 91180
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.309260133745639e-05,
+      "loss": 0.9335,
+      "step": 91190
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3091839688028396e-05,
+      "loss": 0.9347,
+      "step": 91200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.3091078038600394e-05,
+      "loss": 0.9534,
+      "step": 91210
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.309031638917239e-05,
+      "loss": 1.0067,
+      "step": 91220
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308955473974439e-05,
+      "loss": 0.9445,
+      "step": 91230
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308879309031639e-05,
+      "loss": 0.9248,
+      "step": 91240
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308803144088839e-05,
+      "loss": 1.0206,
+      "step": 91250
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308726979146039e-05,
+      "loss": 0.9477,
+      "step": 91260
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308650814203239e-05,
+      "loss": 0.9507,
+      "step": 91270
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308574649260438e-05,
+      "loss": 1.0352,
+      "step": 91280
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308498484317639e-05,
+      "loss": 1.0898,
+      "step": 91290
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3084223193748385e-05,
+      "loss": 1.0044,
+      "step": 91300
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308346154432038e-05,
+      "loss": 0.9576,
+      "step": 91310
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308269989489238e-05,
+      "loss": 0.9163,
+      "step": 91320
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3081938245464384e-05,
+      "loss": 1.0287,
+      "step": 91330
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3081176596036374e-05,
+      "loss": 1.007,
+      "step": 91340
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.308041494660838e-05,
+      "loss": 0.9235,
+      "step": 91350
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3079653297180376e-05,
+      "loss": 0.9923,
+      "step": 91360
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3078891647752374e-05,
+      "loss": 1.0594,
+      "step": 91370
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.307812999832437e-05,
+      "loss": 0.882,
+      "step": 91380
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3077368348896375e-05,
+      "loss": 1.0568,
+      "step": 91390
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3076606699468366e-05,
+      "loss": 0.8676,
+      "step": 91400
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.307584505004037e-05,
+      "loss": 0.9238,
+      "step": 91410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.307508340061237e-05,
+      "loss": 1.0302,
+      "step": 91420
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3074321751184365e-05,
+      "loss": 0.9627,
+      "step": 91430
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.307356010175637e-05,
+      "loss": 1.077,
+      "step": 91440
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.307279845232837e-05,
+      "loss": 0.9092,
+      "step": 91450
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3072036802900364e-05,
+      "loss": 1.0114,
+      "step": 91460
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.307127515347236e-05,
+      "loss": 1.0952,
+      "step": 91470
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3070513504044366e-05,
+      "loss": 0.9451,
+      "step": 91480
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306975185461636e-05,
+      "loss": 1.0384,
+      "step": 91490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306899020518836e-05,
+      "loss": 1.0418,
+      "step": 91500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306822855576036e-05,
+      "loss": 0.96,
+      "step": 91510
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3067466906332356e-05,
+      "loss": 0.9897,
+      "step": 91520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3066705256904353e-05,
+      "loss": 0.9894,
+      "step": 91530
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306594360747636e-05,
+      "loss": 1.013,
+      "step": 91540
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306518195804835e-05,
+      "loss": 1.067,
+      "step": 91550
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306442030862035e-05,
+      "loss": 1.0646,
+      "step": 91560
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306365865919234e-05,
+      "loss": 1.1003,
+      "step": 91570
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306289700976435e-05,
+      "loss": 1.089,
+      "step": 91580
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3062135360336345e-05,
+      "loss": 1.0082,
+      "step": 91590
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306137371090834e-05,
+      "loss": 0.9634,
+      "step": 91600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.306061206148034e-05,
+      "loss": 0.9643,
+      "step": 91610
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3059850412052344e-05,
+      "loss": 1.0255,
+      "step": 91620
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.305908876262434e-05,
+      "loss": 1.1361,
+      "step": 91630
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.305832711319634e-05,
+      "loss": 1.0282,
+      "step": 91640
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.3057565463768343e-05,
+      "loss": 1.0056,
+      "step": 91650
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3056803814340334e-05,
+      "loss": 0.9191,
+      "step": 91660
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.305604216491234e-05,
+      "loss": 0.9532,
+      "step": 91670
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3055280515484336e-05,
+      "loss": 0.9384,
+      "step": 91680
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.305451886605633e-05,
+      "loss": 0.9061,
+      "step": 91690
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.305375721662833e-05,
+      "loss": 0.9894,
+      "step": 91700
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3052995567200335e-05,
+      "loss": 1.0887,
+      "step": 91710
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3052233917772326e-05,
+      "loss": 1.1651,
+      "step": 91720
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.305147226834433e-05,
+      "loss": 0.9978,
+      "step": 91730
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.305071061891633e-05,
+      "loss": 1.0495,
+      "step": 91740
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3049948969488325e-05,
+      "loss": 0.9006,
+      "step": 91750
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304918732006032e-05,
+      "loss": 0.8559,
+      "step": 91760
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3048425670632327e-05,
+      "loss": 1.1485,
+      "step": 91770
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304766402120432e-05,
+      "loss": 0.9426,
+      "step": 91780
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304690237177632e-05,
+      "loss": 1.001,
+      "step": 91790
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304614072234832e-05,
+      "loss": 0.9335,
+      "step": 91800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3045379072920316e-05,
+      "loss": 0.9344,
+      "step": 91810
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3044617423492314e-05,
+      "loss": 1.119,
+      "step": 91820
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304385577406432e-05,
+      "loss": 1.105,
+      "step": 91830
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3043094124636316e-05,
+      "loss": 0.9653,
+      "step": 91840
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304233247520831e-05,
+      "loss": 1.1216,
+      "step": 91850
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.304157082578032e-05,
+      "loss": 1.1377,
+      "step": 91860
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.304080917635231e-05,
+      "loss": 0.9282,
+      "step": 91870
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.304004752692431e-05,
+      "loss": 0.9686,
+      "step": 91880
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.303928587749631e-05,
+      "loss": 1.0684,
+      "step": 91890
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.303852422806831e-05,
+      "loss": 1.0132,
+      "step": 91900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3037762578640305e-05,
+      "loss": 1.042,
+      "step": 91910
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.303700092921231e-05,
+      "loss": 0.9713,
+      "step": 91920
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.30362392797843e-05,
+      "loss": 1.0081,
+      "step": 91930
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3035477630356304e-05,
+      "loss": 1.0939,
+      "step": 91940
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.30347159809283e-05,
+      "loss": 1.1411,
+      "step": 91950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.30339543315003e-05,
+      "loss": 0.9818,
+      "step": 91960
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3033192682072296e-05,
+      "loss": 1.0112,
+      "step": 91970
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.30324310326443e-05,
+      "loss": 1.0149,
+      "step": 91980
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.303166938321629e-05,
+      "loss": 1.0319,
+      "step": 91990
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3030907733788295e-05,
+      "loss": 0.8882,
+      "step": 92000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.303014608436029e-05,
+      "loss": 0.9191,
+      "step": 92010
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.302938443493229e-05,
+      "loss": 1.0173,
+      "step": 92020
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.302862278550429e-05,
+      "loss": 0.9551,
+      "step": 92030
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3027861136076285e-05,
+      "loss": 0.9622,
+      "step": 92040
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.302709948664829e-05,
+      "loss": 1.1301,
+      "step": 92050
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.302633783722029e-05,
+      "loss": 1.0336,
+      "step": 92060
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3025576187792284e-05,
+      "loss": 1.0421,
+      "step": 92070
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.302481453836428e-05,
+      "loss": 1.084,
+      "step": 92080
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3024052888936286e-05,
+      "loss": 0.957,
+      "step": 92090
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.302329123950828e-05,
+      "loss": 1.0499,
+      "step": 92100
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.302252959008028e-05,
+      "loss": 0.9711,
+      "step": 92110
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.302176794065228e-05,
+      "loss": 0.9954,
+      "step": 92120
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3021006291224276e-05,
+      "loss": 0.9987,
+      "step": 92130
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3020244641796273e-05,
+      "loss": 1.0363,
+      "step": 92140
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301948299236828e-05,
+      "loss": 0.9525,
+      "step": 92150
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301872134294027e-05,
+      "loss": 1.0464,
+      "step": 92160
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301795969351227e-05,
+      "loss": 0.9096,
+      "step": 92170
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301719804408427e-05,
+      "loss": 1.1093,
+      "step": 92180
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301643639465627e-05,
+      "loss": 1.0113,
+      "step": 92190
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3015674745228265e-05,
+      "loss": 0.983,
+      "step": 92200
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301491309580027e-05,
+      "loss": 0.9177,
+      "step": 92210
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301415144637227e-05,
+      "loss": 1.0371,
+      "step": 92220
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3013389796944264e-05,
+      "loss": 0.9587,
+      "step": 92230
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301262814751627e-05,
+      "loss": 1.0689,
+      "step": 92240
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301186649808826e-05,
+      "loss": 1.0668,
+      "step": 92250
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3011104848660263e-05,
+      "loss": 1.0317,
+      "step": 92260
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.301034319923226e-05,
+      "loss": 0.9869,
+      "step": 92270
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.300958154980426e-05,
+      "loss": 1.2368,
+      "step": 92280
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3008819900376256e-05,
+      "loss": 1.1293,
+      "step": 92290
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.300805825094826e-05,
+      "loss": 0.9413,
+      "step": 92300
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300729660152025e-05,
+      "loss": 1.0353,
+      "step": 92310
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.3006534952092255e-05,
+      "loss": 1.0263,
+      "step": 92320
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300577330266425e-05,
+      "loss": 1.0564,
+      "step": 92330
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300501165323625e-05,
+      "loss": 1.066,
+      "step": 92340
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300425000380825e-05,
+      "loss": 1.0996,
+      "step": 92350
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300348835438025e-05,
+      "loss": 1.0726,
+      "step": 92360
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300272670495224e-05,
+      "loss": 1.0143,
+      "step": 92370
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.3001965055524247e-05,
+      "loss": 0.8976,
+      "step": 92380
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.3001203406096244e-05,
+      "loss": 0.9993,
+      "step": 92390
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.300044175666824e-05,
+      "loss": 0.885,
+      "step": 92400
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299968010724024e-05,
+      "loss": 1.0505,
+      "step": 92410
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299891845781224e-05,
+      "loss": 0.9403,
+      "step": 92420
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299815680838424e-05,
+      "loss": 1.048,
+      "step": 92430
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299739515895624e-05,
+      "loss": 1.1297,
+      "step": 92440
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299663350952824e-05,
+      "loss": 0.9682,
+      "step": 92450
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299587186010023e-05,
+      "loss": 0.9537,
+      "step": 92460
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299511021067224e-05,
+      "loss": 1.0009,
+      "step": 92470
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299434856124423e-05,
+      "loss": 1.0555,
+      "step": 92480
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299358691181623e-05,
+      "loss": 0.9663,
+      "step": 92490
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299282526238823e-05,
+      "loss": 1.0341,
+      "step": 92500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299206361296023e-05,
+      "loss": 1.0758,
+      "step": 92510
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2991301963532225e-05,
+      "loss": 1.0985,
+      "step": 92520
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.299054031410423e-05,
+      "loss": 1.1301,
+      "step": 92530
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298977866467622e-05,
+      "loss": 1.029,
+      "step": 92540
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2989017015248224e-05,
+      "loss": 1.1523,
+      "step": 92550
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298825536582022e-05,
+      "loss": 0.9991,
+      "step": 92560
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298749371639222e-05,
+      "loss": 0.8826,
+      "step": 92570
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2986732066964216e-05,
+      "loss": 0.9103,
+      "step": 92580
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298597041753622e-05,
+      "loss": 0.9852,
+      "step": 92590
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298520876810822e-05,
+      "loss": 1.1325,
+      "step": 92600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2984447118680215e-05,
+      "loss": 1.0925,
+      "step": 92610
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298368546925221e-05,
+      "loss": 0.9065,
+      "step": 92620
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298292381982421e-05,
+      "loss": 1.0424,
+      "step": 92630
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2982162170396215e-05,
+      "loss": 1.0605,
+      "step": 92640
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298140052096821e-05,
+      "loss": 1.018,
+      "step": 92650
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.298063887154021e-05,
+      "loss": 1.0098,
+      "step": 92660
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.297987722211221e-05,
+      "loss": 1.0004,
+      "step": 92670
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.297911557268421e-05,
+      "loss": 0.9468,
+      "step": 92680
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.29783539232562e-05,
+      "loss": 0.9876,
+      "step": 92690
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2977592273828206e-05,
+      "loss": 1.0605,
+      "step": 92700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2976830624400204e-05,
+      "loss": 0.8931,
+      "step": 92710
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.29760689749722e-05,
+      "loss": 0.9868,
+      "step": 92720
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.29753073255442e-05,
+      "loss": 1.0482,
+      "step": 92730
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.29745456761162e-05,
+      "loss": 1.0825,
+      "step": 92740
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2973784026688193e-05,
+      "loss": 0.9741,
+      "step": 92750
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.29730223772602e-05,
+      "loss": 0.9949,
+      "step": 92760
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2972260727832195e-05,
+      "loss": 1.0189,
+      "step": 92770
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.297149907840419e-05,
+      "loss": 1.061,
+      "step": 92780
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.297073742897619e-05,
+      "loss": 1.0059,
+      "step": 92790
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2969975779548194e-05,
+      "loss": 1.0647,
+      "step": 92800
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296921413012019e-05,
+      "loss": 1.1236,
+      "step": 92810
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296845248069219e-05,
+      "loss": 0.8988,
+      "step": 92820
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296769083126419e-05,
+      "loss": 0.9859,
+      "step": 92830
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2966929181836184e-05,
+      "loss": 1.074,
+      "step": 92840
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296616753240819e-05,
+      "loss": 0.9627,
+      "step": 92850
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2965405882980186e-05,
+      "loss": 0.9725,
+      "step": 92860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2964644233552183e-05,
+      "loss": 1.0424,
+      "step": 92870
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296388258412418e-05,
+      "loss": 0.8954,
+      "step": 92880
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2963120934696185e-05,
+      "loss": 0.9588,
+      "step": 92890
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2962359285268176e-05,
+      "loss": 1.0373,
+      "step": 92900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296159763584018e-05,
+      "loss": 1.0525,
+      "step": 92910
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.296083598641218e-05,
+      "loss": 0.9369,
+      "step": 92920
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2960074336984175e-05,
+      "loss": 1.0504,
+      "step": 92930
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.295931268755617e-05,
+      "loss": 1.0773,
+      "step": 92940
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.295855103812817e-05,
+      "loss": 0.9279,
+      "step": 92950
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.295778938870017e-05,
+      "loss": 0.9395,
+      "step": 92960
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295702773927217e-05,
+      "loss": 1.1406,
+      "step": 92970
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295626608984417e-05,
+      "loss": 0.9604,
+      "step": 92980
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2955504440416167e-05,
+      "loss": 0.9295,
+      "step": 92990
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2954742790988164e-05,
+      "loss": 1.043,
+      "step": 93000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295398114156016e-05,
+      "loss": 0.9461,
+      "step": 93010
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2953219492132166e-05,
+      "loss": 1.0188,
+      "step": 93020
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295245784270416e-05,
+      "loss": 0.9376,
+      "step": 93030
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295169619327616e-05,
+      "loss": 0.9176,
+      "step": 93040
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295093454384816e-05,
+      "loss": 1.0981,
+      "step": 93050
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.295017289442016e-05,
+      "loss": 1.0036,
+      "step": 93060
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.294941124499215e-05,
+      "loss": 0.9375,
+      "step": 93070
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.294864959556416e-05,
+      "loss": 1.0086,
+      "step": 93080
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2947887946136155e-05,
+      "loss": 0.9745,
+      "step": 93090
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.294712629670815e-05,
+      "loss": 1.1571,
+      "step": 93100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.294636464728015e-05,
+      "loss": 1.0607,
+      "step": 93110
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2945602997852154e-05,
+      "loss": 1.0528,
+      "step": 93120
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2944841348424145e-05,
+      "loss": 0.9493,
+      "step": 93130
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.294407969899615e-05,
+      "loss": 1.0464,
+      "step": 93140
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2943318049568146e-05,
+      "loss": 1.066,
+      "step": 93150
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2942556400140144e-05,
+      "loss": 0.9959,
+      "step": 93160
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.294179475071214e-05,
+      "loss": 0.9697,
+      "step": 93170
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2941033101284146e-05,
+      "loss": 1.0598,
+      "step": 93180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.294027145185614e-05,
+      "loss": 0.9488,
+      "step": 93190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293950980242814e-05,
+      "loss": 0.8812,
+      "step": 93200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293874815300014e-05,
+      "loss": 1.0282,
+      "step": 93210
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2937986503572135e-05,
+      "loss": 0.9658,
+      "step": 93220
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293722485414414e-05,
+      "loss": 0.9361,
+      "step": 93230
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293646320471614e-05,
+      "loss": 0.9555,
+      "step": 93240
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2935701555288135e-05,
+      "loss": 0.9203,
+      "step": 93250
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293493990586013e-05,
+      "loss": 0.9542,
+      "step": 93260
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2934178256432136e-05,
+      "loss": 1.1236,
+      "step": 93270
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293341660700413e-05,
+      "loss": 0.9159,
+      "step": 93280
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293265495757613e-05,
+      "loss": 0.8856,
+      "step": 93290
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.293189330814813e-05,
+      "loss": 1.0349,
+      "step": 93300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2931131658720126e-05,
+      "loss": 0.9705,
+      "step": 93310
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2930370009292124e-05,
+      "loss": 1.0362,
+      "step": 93320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292960835986413e-05,
+      "loss": 1.0431,
+      "step": 93330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292884671043612e-05,
+      "loss": 0.9868,
+      "step": 93340
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292808506100812e-05,
+      "loss": 1.032,
+      "step": 93350
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292732341158012e-05,
+      "loss": 0.9099,
+      "step": 93360
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292656176215212e-05,
+      "loss": 1.013,
+      "step": 93370
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2925800112724115e-05,
+      "loss": 1.0758,
+      "step": 93380
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292503846329612e-05,
+      "loss": 0.9754,
+      "step": 93390
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.292427681386812e-05,
+      "loss": 0.9349,
+      "step": 93400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2923515164440114e-05,
+      "loss": 0.9543,
+      "step": 93410
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.292275351501211e-05,
+      "loss": 0.8543,
+      "step": 93420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.292199186558411e-05,
+      "loss": 0.9733,
+      "step": 93430
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2921230216156114e-05,
+      "loss": 1.083,
+      "step": 93440
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2920468566728104e-05,
+      "loss": 1.0741,
+      "step": 93450
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.291970691730011e-05,
+      "loss": 1.0784,
+      "step": 93460
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2918945267872106e-05,
+      "loss": 1.0088,
+      "step": 93470
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2918183618444103e-05,
+      "loss": 1.0836,
+      "step": 93480
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.29174219690161e-05,
+      "loss": 0.9823,
+      "step": 93490
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2916660319588105e-05,
+      "loss": 1.0483,
+      "step": 93500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2915898670160096e-05,
+      "loss": 0.9735,
+      "step": 93510
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.29151370207321e-05,
+      "loss": 0.9366,
+      "step": 93520
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.29143753713041e-05,
+      "loss": 0.8982,
+      "step": 93530
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2913613721876095e-05,
+      "loss": 1.013,
+      "step": 93540
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.291285207244809e-05,
+      "loss": 1.0086,
+      "step": 93550
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.29120904230201e-05,
+      "loss": 1.0373,
+      "step": 93560
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2911328773592094e-05,
+      "loss": 1.1475,
+      "step": 93570
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.291056712416409e-05,
+      "loss": 0.9123,
+      "step": 93580
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.290980547473609e-05,
+      "loss": 0.9156,
+      "step": 93590
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2909043825308087e-05,
+      "loss": 1.014,
+      "step": 93600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.290828217588009e-05,
+      "loss": 1.0891,
+      "step": 93610
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.290752052645209e-05,
+      "loss": 0.958,
+      "step": 93620
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2906758877024086e-05,
+      "loss": 0.9817,
+      "step": 93630
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290599722759608e-05,
+      "loss": 0.9622,
+      "step": 93640
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290523557816809e-05,
+      "loss": 0.9167,
+      "step": 93650
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290447392874008e-05,
+      "loss": 0.891,
+      "step": 93660
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290371227931208e-05,
+      "loss": 0.9654,
+      "step": 93670
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290295062988408e-05,
+      "loss": 1.0245,
+      "step": 93680
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290218898045608e-05,
+      "loss": 0.8696,
+      "step": 93690
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2901427331028075e-05,
+      "loss": 0.9424,
+      "step": 93700
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.290066568160008e-05,
+      "loss": 0.9709,
+      "step": 93710
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289990403217207e-05,
+      "loss": 1.0893,
+      "step": 93720
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2899142382744074e-05,
+      "loss": 1.0099,
+      "step": 93730
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289838073331607e-05,
+      "loss": 0.9669,
+      "step": 93740
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289761908388807e-05,
+      "loss": 0.8656,
+      "step": 93750
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2896857434460066e-05,
+      "loss": 1.0239,
+      "step": 93760
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289609578503207e-05,
+      "loss": 1.184,
+      "step": 93770
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289533413560407e-05,
+      "loss": 0.9705,
+      "step": 93780
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2894572486176066e-05,
+      "loss": 1.0865,
+      "step": 93790
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289381083674806e-05,
+      "loss": 1.1063,
+      "step": 93800
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289304918732006e-05,
+      "loss": 1.205,
+      "step": 93810
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2892287537892065e-05,
+      "loss": 1.0424,
+      "step": 93820
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289152588846406e-05,
+      "loss": 1.0103,
+      "step": 93830
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.289076423903606e-05,
+      "loss": 0.9067,
+      "step": 93840
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.289000258960806e-05,
+      "loss": 1.0209,
+      "step": 93850
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2889240940180055e-05,
+      "loss": 1.2228,
+      "step": 93860
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288847929075205e-05,
+      "loss": 1.0733,
+      "step": 93870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2887717641324056e-05,
+      "loss": 0.9885,
+      "step": 93880
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288695599189605e-05,
+      "loss": 0.918,
+      "step": 93890
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288619434246805e-05,
+      "loss": 0.9354,
+      "step": 93900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288543269304005e-05,
+      "loss": 0.9443,
+      "step": 93910
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2884671043612046e-05,
+      "loss": 0.9373,
+      "step": 93920
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2883909394184044e-05,
+      "loss": 1.0552,
+      "step": 93930
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288314774475605e-05,
+      "loss": 1.1145,
+      "step": 93940
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2882386095328045e-05,
+      "loss": 0.9385,
+      "step": 93950
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288162444590004e-05,
+      "loss": 1.0952,
+      "step": 93960
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288086279647204e-05,
+      "loss": 0.9961,
+      "step": 93970
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.288010114704404e-05,
+      "loss": 1.0586,
+      "step": 93980
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.287933949761604e-05,
+      "loss": 0.8887,
+      "step": 93990
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.287857784818804e-05,
+      "loss": 0.9727,
+      "step": 94000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.287781619876004e-05,
+      "loss": 1.0256,
+      "step": 94010
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2877054549332034e-05,
+      "loss": 1.0101,
+      "step": 94020
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.287629289990404e-05,
+      "loss": 0.922,
+      "step": 94030
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.287553125047603e-05,
+      "loss": 0.9399,
+      "step": 94040
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2874769601048034e-05,
+      "loss": 1.0303,
+      "step": 94050
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.287400795162003e-05,
+      "loss": 1.0899,
+      "step": 94060
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.287324630219203e-05,
+      "loss": 0.9419,
+      "step": 94070
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2872484652764026e-05,
+      "loss": 0.9732,
+      "step": 94080
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.287172300333603e-05,
+      "loss": 0.9969,
+      "step": 94090
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.287096135390802e-05,
+      "loss": 0.8509,
+      "step": 94100
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2870199704480025e-05,
+      "loss": 0.9918,
+      "step": 94110
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286943805505202e-05,
+      "loss": 0.9606,
+      "step": 94120
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286867640562402e-05,
+      "loss": 0.9799,
+      "step": 94130
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286791475619602e-05,
+      "loss": 1.1403,
+      "step": 94140
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286715310676802e-05,
+      "loss": 0.9512,
+      "step": 94150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286639145734002e-05,
+      "loss": 1.0111,
+      "step": 94160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286562980791202e-05,
+      "loss": 1.054,
+      "step": 94170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2864868158484014e-05,
+      "loss": 0.8694,
+      "step": 94180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286410650905601e-05,
+      "loss": 0.8899,
+      "step": 94190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2863344859628016e-05,
+      "loss": 0.9584,
+      "step": 94200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286258321020001e-05,
+      "loss": 0.9742,
+      "step": 94210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286182156077201e-05,
+      "loss": 1.0179,
+      "step": 94220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286105991134401e-05,
+      "loss": 0.9923,
+      "step": 94230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.286029826191601e-05,
+      "loss": 1.054,
+      "step": 94240
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2859536612488e-05,
+      "loss": 1.0479,
+      "step": 94250
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.285877496306001e-05,
+      "loss": 1.1018,
+      "step": 94260
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2858013313632005e-05,
+      "loss": 0.9079,
+      "step": 94270
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2857251664204e-05,
+      "loss": 1.0551,
+      "step": 94280
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2856490014776e-05,
+      "loss": 0.8973,
+      "step": 94290
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2855728365348004e-05,
+      "loss": 0.9762,
+      "step": 94300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2854966715919995e-05,
+      "loss": 1.0369,
+      "step": 94310
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2854205066492e-05,
+      "loss": 0.9739,
+      "step": 94320
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.285344341706399e-05,
+      "loss": 1.05,
+      "step": 94330
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2852681767635994e-05,
+      "loss": 1.0046,
+      "step": 94340
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.285192011820799e-05,
+      "loss": 0.8821,
+      "step": 94350
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.285115846877999e-05,
+      "loss": 1.1369,
+      "step": 94360
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.285039681935199e-05,
+      "loss": 1.0503,
+      "step": 94370
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284963516992399e-05,
+      "loss": 0.9834,
+      "step": 94380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284887352049599e-05,
+      "loss": 0.9398,
+      "step": 94390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2848111871067986e-05,
+      "loss": 0.9992,
+      "step": 94400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284735022163999e-05,
+      "loss": 0.9989,
+      "step": 94410
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284658857221198e-05,
+      "loss": 0.9856,
+      "step": 94420
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2845826922783985e-05,
+      "loss": 0.9641,
+      "step": 94430
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284506527335598e-05,
+      "loss": 0.8841,
+      "step": 94440
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284430362392798e-05,
+      "loss": 1.1581,
+      "step": 94450
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284354197449998e-05,
+      "loss": 0.9159,
+      "step": 94460
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284278032507198e-05,
+      "loss": 1.0508,
+      "step": 94470
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.284201867564397e-05,
+      "loss": 1.0127,
+      "step": 94480
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.2841257026215976e-05,
+      "loss": 0.9775,
+      "step": 94490
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2840495376787974e-05,
+      "loss": 1.0227,
+      "step": 94500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283973372735997e-05,
+      "loss": 1.0299,
+      "step": 94510
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283897207793197e-05,
+      "loss": 0.958,
+      "step": 94520
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283821042850397e-05,
+      "loss": 1.0166,
+      "step": 94530
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283744877907597e-05,
+      "loss": 0.9152,
+      "step": 94540
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283668712964797e-05,
+      "loss": 1.1527,
+      "step": 94550
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2835925480219965e-05,
+      "loss": 1.0699,
+      "step": 94560
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283516383079196e-05,
+      "loss": 0.8806,
+      "step": 94570
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283440218136397e-05,
+      "loss": 0.9802,
+      "step": 94580
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2833640531935965e-05,
+      "loss": 1.0282,
+      "step": 94590
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283287888250796e-05,
+      "loss": 0.9392,
+      "step": 94600
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.283211723307996e-05,
+      "loss": 1.0508,
+      "step": 94610
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2831355583651964e-05,
+      "loss": 0.9292,
+      "step": 94620
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2830593934223954e-05,
+      "loss": 0.9565,
+      "step": 94630
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.282983228479596e-05,
+      "loss": 1.0303,
+      "step": 94640
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2829070635367956e-05,
+      "loss": 1.0276,
+      "step": 94650
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2828308985939954e-05,
+      "loss": 0.9725,
+      "step": 94660
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.282754733651195e-05,
+      "loss": 1.0508,
+      "step": 94670
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2826785687083955e-05,
+      "loss": 1.1064,
+      "step": 94680
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.2826024037655946e-05,
+      "loss": 1.0037,
+      "step": 94690
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.282526238822795e-05,
+      "loss": 1.0311,
+      "step": 94700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.282450073879995e-05,
+      "loss": 0.9709,
+      "step": 94710
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.2823739089371945e-05,
+      "loss": 1.009,
+      "step": 94720
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.282297743994394e-05,
+      "loss": 1.1876,
+      "step": 94730
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.282221579051595e-05,
+      "loss": 0.9899,
+      "step": 94740
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.2821454141087944e-05,
+      "loss": 1.0717,
+      "step": 94750
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.282069249165994e-05,
+      "loss": 1.0449,
+      "step": 94760
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281993084223194e-05,
+      "loss": 0.9997,
+      "step": 94770
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281916919280394e-05,
+      "loss": 1.2418,
+      "step": 94780
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281840754337594e-05,
+      "loss": 1.0073,
+      "step": 94790
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281764589394793e-05,
+      "loss": 0.9452,
+      "step": 94800
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.2816884244519936e-05,
+      "loss": 1.0767,
+      "step": 94810
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281612259509193e-05,
+      "loss": 0.9792,
+      "step": 94820
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281536094566393e-05,
+      "loss": 0.8542,
+      "step": 94830
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281459929623593e-05,
+      "loss": 0.9498,
+      "step": 94840
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281383764680793e-05,
+      "loss": 0.9867,
+      "step": 94850
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281307599737992e-05,
+      "loss": 1.088,
+      "step": 94860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281231434795193e-05,
+      "loss": 0.9996,
+      "step": 94870
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.2811552698523925e-05,
+      "loss": 1.0202,
+      "step": 94880
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281079104909592e-05,
+      "loss": 0.8977,
+      "step": 94890
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.281002939966792e-05,
+      "loss": 1.0361,
+      "step": 94900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.2809267750239924e-05,
+      "loss": 1.0848,
+      "step": 94910
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.2808506100811915e-05,
+      "loss": 0.9183,
+      "step": 94920
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.280774445138392e-05,
+      "loss": 0.999,
+      "step": 94930
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2806982801955917e-05,
+      "loss": 1.0499,
+      "step": 94940
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2806221152527914e-05,
+      "loss": 0.9737,
+      "step": 94950
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.280545950309992e-05,
+      "loss": 0.9253,
+      "step": 94960
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2804697853671916e-05,
+      "loss": 1.0277,
+      "step": 94970
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.280393620424391e-05,
+      "loss": 0.9663,
+      "step": 94980
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.280317455481591e-05,
+      "loss": 0.9501,
+      "step": 94990
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2802412905387915e-05,
+      "loss": 1.0175,
+      "step": 95000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2801651255959906e-05,
+      "loss": 0.9103,
+      "step": 95010
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.280088960653191e-05,
+      "loss": 1.0114,
+      "step": 95020
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.280012795710391e-05,
+      "loss": 0.9515,
+      "step": 95030
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2799366307675905e-05,
+      "loss": 0.9965,
+      "step": 95040
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.27986046582479e-05,
+      "loss": 0.9786,
+      "step": 95050
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2797843008819906e-05,
+      "loss": 0.9386,
+      "step": 95060
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.27970813593919e-05,
+      "loss": 0.9547,
+      "step": 95070
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.27963197099639e-05,
+      "loss": 0.9379,
+      "step": 95080
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.27955580605359e-05,
+      "loss": 1.1088,
+      "step": 95090
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2794796411107896e-05,
+      "loss": 0.9276,
+      "step": 95100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.2794034761679894e-05,
+      "loss": 1.052,
+      "step": 95110
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.27932731122519e-05,
+      "loss": 0.9086,
+      "step": 95120
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.279251146282389e-05,
+      "loss": 1.0571,
+      "step": 95130
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.279174981339589e-05,
+      "loss": 0.9315,
+      "step": 95140
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.279098816396789e-05,
+      "loss": 0.9392,
+      "step": 95150
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.279022651453989e-05,
+      "loss": 1.0147,
+      "step": 95160
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278946486511189e-05,
+      "loss": 1.1424,
+      "step": 95170
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278870321568389e-05,
+      "loss": 1.138,
+      "step": 95180
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278794156625589e-05,
+      "loss": 1.0052,
+      "step": 95190
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2787179916827885e-05,
+      "loss": 1.0811,
+      "step": 95200
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278641826739989e-05,
+      "loss": 0.9731,
+      "step": 95210
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278565661797188e-05,
+      "loss": 0.937,
+      "step": 95220
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2784894968543884e-05,
+      "loss": 1.0364,
+      "step": 95230
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2784133319115874e-05,
+      "loss": 1.1294,
+      "step": 95240
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278337166968788e-05,
+      "loss": 0.8551,
+      "step": 95250
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2782610020259876e-05,
+      "loss": 1.002,
+      "step": 95260
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2781848370831874e-05,
+      "loss": 0.9913,
+      "step": 95270
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.278108672140387e-05,
+      "loss": 1.012,
+      "step": 95280
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2780325071975875e-05,
+      "loss": 1.0531,
+      "step": 95290
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2779563422547866e-05,
+      "loss": 0.9395,
+      "step": 95300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.277880177311987e-05,
+      "loss": 0.9812,
+      "step": 95310
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.277804012369187e-05,
+      "loss": 1.0276,
+      "step": 95320
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2777278474263865e-05,
+      "loss": 0.9557,
+      "step": 95330
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.277651682483587e-05,
+      "loss": 1.0748,
+      "step": 95340
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.277575517540787e-05,
+      "loss": 0.9438,
+      "step": 95350
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.2774993525979864e-05,
+      "loss": 0.9715,
+      "step": 95360
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.277423187655186e-05,
+      "loss": 1.0652,
+      "step": 95370
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2773470227123866e-05,
+      "loss": 1.0615,
+      "step": 95380
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.277270857769586e-05,
+      "loss": 0.9927,
+      "step": 95390
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.277194692826786e-05,
+      "loss": 1.0401,
+      "step": 95400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.277118527883986e-05,
+      "loss": 1.0528,
+      "step": 95410
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2770423629411856e-05,
+      "loss": 0.9687,
+      "step": 95420
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2769661979983853e-05,
+      "loss": 1.0094,
+      "step": 95430
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276890033055586e-05,
+      "loss": 1.0324,
+      "step": 95440
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276813868112785e-05,
+      "loss": 0.9751,
+      "step": 95450
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276737703169985e-05,
+      "loss": 1.0446,
+      "step": 95460
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276661538227185e-05,
+      "loss": 0.981,
+      "step": 95470
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276585373284385e-05,
+      "loss": 1.0161,
+      "step": 95480
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2765092083415845e-05,
+      "loss": 0.9485,
+      "step": 95490
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276433043398785e-05,
+      "loss": 0.9991,
+      "step": 95500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276356878455984e-05,
+      "loss": 0.9141,
+      "step": 95510
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2762807135131844e-05,
+      "loss": 0.9914,
+      "step": 95520
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276204548570384e-05,
+      "loss": 1.1075,
+      "step": 95530
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276128383627584e-05,
+      "loss": 0.8898,
+      "step": 95540
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.276052218684784e-05,
+      "loss": 1.1121,
+      "step": 95550
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.275976053741984e-05,
+      "loss": 0.9695,
+      "step": 95560
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.275899888799184e-05,
+      "loss": 1.0255,
+      "step": 95570
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2758237238563836e-05,
+      "loss": 0.9954,
+      "step": 95580
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.275747558913584e-05,
+      "loss": 1.0766,
+      "step": 95590
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.275671393970783e-05,
+      "loss": 0.9785,
+      "step": 95600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2755952290279835e-05,
+      "loss": 1.0021,
+      "step": 95610
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.275519064085183e-05,
+      "loss": 1.1045,
+      "step": 95620
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.275442899142383e-05,
+      "loss": 0.9607,
+      "step": 95630
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.275366734199583e-05,
+      "loss": 1.0116,
+      "step": 95640
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.275290569256783e-05,
+      "loss": 0.9268,
+      "step": 95650
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.275214404313982e-05,
+      "loss": 1.1055,
+      "step": 95660
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2751382393711826e-05,
+      "loss": 0.9897,
+      "step": 95670
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2750620744283824e-05,
+      "loss": 0.9505,
+      "step": 95680
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274985909485582e-05,
+      "loss": 1.0512,
+      "step": 95690
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274909744542782e-05,
+      "loss": 0.9606,
+      "step": 95700
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2748335795999816e-05,
+      "loss": 1.0162,
+      "step": 95710
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2747574146571814e-05,
+      "loss": 0.9899,
+      "step": 95720
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274681249714382e-05,
+      "loss": 0.9203,
+      "step": 95730
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2746050847715816e-05,
+      "loss": 0.927,
+      "step": 95740
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274528919828781e-05,
+      "loss": 1.0142,
+      "step": 95750
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274452754885982e-05,
+      "loss": 1.0508,
+      "step": 95760
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274376589943181e-05,
+      "loss": 1.2082,
+      "step": 95770
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274300425000381e-05,
+      "loss": 1.1411,
+      "step": 95780
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274224260057581e-05,
+      "loss": 1.0079,
+      "step": 95790
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.274148095114781e-05,
+      "loss": 0.8953,
+      "step": 95800
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.2740719301719805e-05,
+      "loss": 1.0653,
+      "step": 95810
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.273995765229181e-05,
+      "loss": 0.9993,
+      "step": 95820
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.27391960028638e-05,
+      "loss": 1.0409,
+      "step": 95830
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.2738434353435804e-05,
+      "loss": 1.0153,
+      "step": 95840
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.27376727040078e-05,
+      "loss": 0.9725,
+      "step": 95850
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.27369110545798e-05,
+      "loss": 1.0225,
+      "step": 95860
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.2736149405151796e-05,
+      "loss": 0.9559,
+      "step": 95870
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.27353877557238e-05,
+      "loss": 0.9376,
+      "step": 95880
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.273462610629579e-05,
+      "loss": 0.8804,
+      "step": 95890
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.2733864456867795e-05,
+      "loss": 1.2108,
+      "step": 95900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.273310280743979e-05,
+      "loss": 1.0014,
+      "step": 95910
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.273234115801179e-05,
+      "loss": 1.0151,
+      "step": 95920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.2731579508583794e-05,
+      "loss": 0.9645,
+      "step": 95930
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.273081785915579e-05,
+      "loss": 1.094,
+      "step": 95940
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.273005620972779e-05,
+      "loss": 0.947,
+      "step": 95950
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.272929456029979e-05,
+      "loss": 1.2329,
+      "step": 95960
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.272853291087179e-05,
+      "loss": 1.1818,
+      "step": 95970
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.272777126144378e-05,
+      "loss": 1.1056,
+      "step": 95980
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.2727009612015786e-05,
+      "loss": 0.9995,
+      "step": 95990
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.2726247962587784e-05,
+      "loss": 0.9306,
+      "step": 96000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.272548631315978e-05,
+      "loss": 0.9791,
+      "step": 96010
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.272472466373178e-05,
+      "loss": 1.1141,
+      "step": 96020
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.272396301430378e-05,
+      "loss": 1.0309,
+      "step": 96030
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2723201364875773e-05,
+      "loss": 1.006,
+      "step": 96040
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.272243971544778e-05,
+      "loss": 1.1757,
+      "step": 96050
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2721678066019775e-05,
+      "loss": 0.9914,
+      "step": 96060
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.272091641659177e-05,
+      "loss": 1.1316,
+      "step": 96070
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.272015476716377e-05,
+      "loss": 1.1125,
+      "step": 96080
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2719393117735774e-05,
+      "loss": 1.0844,
+      "step": 96090
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2718631468307765e-05,
+      "loss": 1.1002,
+      "step": 96100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271786981887977e-05,
+      "loss": 1.1333,
+      "step": 96110
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271710816945177e-05,
+      "loss": 1.02,
+      "step": 96120
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2716346520023764e-05,
+      "loss": 0.9551,
+      "step": 96130
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271558487059577e-05,
+      "loss": 0.9717,
+      "step": 96140
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2714823221167766e-05,
+      "loss": 0.9896,
+      "step": 96150
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271406157173976e-05,
+      "loss": 0.9419,
+      "step": 96160
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271329992231176e-05,
+      "loss": 0.905,
+      "step": 96170
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271253827288376e-05,
+      "loss": 0.8826,
+      "step": 96180
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2711776623455756e-05,
+      "loss": 0.9985,
+      "step": 96190
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271101497402776e-05,
+      "loss": 0.961,
+      "step": 96200
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.271025332459975e-05,
+      "loss": 1.1269,
+      "step": 96210
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.2709491675171755e-05,
+      "loss": 0.9984,
+      "step": 96220
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.270873002574375e-05,
+      "loss": 1.0018,
+      "step": 96230
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.270796837631575e-05,
+      "loss": 0.9885,
+      "step": 96240
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270720672688775e-05,
+      "loss": 1.1776,
+      "step": 96250
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270644507745975e-05,
+      "loss": 1.0277,
+      "step": 96260
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270568342803174e-05,
+      "loss": 0.9728,
+      "step": 96270
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2704921778603746e-05,
+      "loss": 1.0688,
+      "step": 96280
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2704160129175744e-05,
+      "loss": 0.9952,
+      "step": 96290
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270339847974774e-05,
+      "loss": 0.8759,
+      "step": 96300
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270263683031974e-05,
+      "loss": 1.0747,
+      "step": 96310
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270187518089174e-05,
+      "loss": 1.0447,
+      "step": 96320
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270111353146374e-05,
+      "loss": 0.9562,
+      "step": 96330
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.270035188203574e-05,
+      "loss": 0.9713,
+      "step": 96340
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269959023260774e-05,
+      "loss": 1.0273,
+      "step": 96350
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269882858317973e-05,
+      "loss": 0.967,
+      "step": 96360
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269806693375174e-05,
+      "loss": 0.975,
+      "step": 96370
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2697305284323735e-05,
+      "loss": 0.986,
+      "step": 96380
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269654363489573e-05,
+      "loss": 0.9872,
+      "step": 96390
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269578198546773e-05,
+      "loss": 0.8769,
+      "step": 96400
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2695020336039734e-05,
+      "loss": 1.0357,
+      "step": 96410
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2694258686611725e-05,
+      "loss": 1.0074,
+      "step": 96420
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269349703718373e-05,
+      "loss": 0.8899,
+      "step": 96430
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2692735387755726e-05,
+      "loss": 1.0438,
+      "step": 96440
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.2691973738327724e-05,
+      "loss": 0.9714,
+      "step": 96450
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.269121208889972e-05,
+      "loss": 1.0082,
+      "step": 96460
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2690450439471725e-05,
+      "loss": 1.0546,
+      "step": 96470
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2689688790043716e-05,
+      "loss": 0.9473,
+      "step": 96480
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268892714061572e-05,
+      "loss": 1.0493,
+      "step": 96490
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268816549118772e-05,
+      "loss": 1.0,
+      "step": 96500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2687403841759715e-05,
+      "loss": 1.1244,
+      "step": 96510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268664219233171e-05,
+      "loss": 1.0435,
+      "step": 96520
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268588054290372e-05,
+      "loss": 0.9397,
+      "step": 96530
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2685118893475715e-05,
+      "loss": 0.9634,
+      "step": 96540
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268435724404771e-05,
+      "loss": 0.978,
+      "step": 96550
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2683595594619716e-05,
+      "loss": 1.0397,
+      "step": 96560
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268283394519171e-05,
+      "loss": 0.9858,
+      "step": 96570
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268207229576371e-05,
+      "loss": 1.0292,
+      "step": 96580
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.268131064633571e-05,
+      "loss": 1.0591,
+      "step": 96590
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2680548996907706e-05,
+      "loss": 1.0114,
+      "step": 96600
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2679787347479704e-05,
+      "loss": 0.9974,
+      "step": 96610
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.26790256980517e-05,
+      "loss": 0.8928,
+      "step": 96620
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.26782640486237e-05,
+      "loss": 1.0173,
+      "step": 96630
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.26775023991957e-05,
+      "loss": 0.906,
+      "step": 96640
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2676740749767693e-05,
+      "loss": 0.9962,
+      "step": 96650
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.26759791003397e-05,
+      "loss": 0.9688,
+      "step": 96660
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.2675217450911695e-05,
+      "loss": 1.0183,
+      "step": 96670
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.267445580148369e-05,
+      "loss": 0.9957,
+      "step": 96680
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.267369415205569e-05,
+      "loss": 0.9639,
+      "step": 96690
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2672932502627694e-05,
+      "loss": 1.027,
+      "step": 96700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.267217085319969e-05,
+      "loss": 0.9762,
+      "step": 96710
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.267140920377169e-05,
+      "loss": 1.0067,
+      "step": 96720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2670647554343693e-05,
+      "loss": 1.0356,
+      "step": 96730
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2669885904915684e-05,
+      "loss": 1.0817,
+      "step": 96740
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266912425548769e-05,
+      "loss": 0.9487,
+      "step": 96750
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2668362606059686e-05,
+      "loss": 0.9884,
+      "step": 96760
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266760095663168e-05,
+      "loss": 0.918,
+      "step": 96770
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266683930720368e-05,
+      "loss": 0.9919,
+      "step": 96780
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2666077657775685e-05,
+      "loss": 1.0858,
+      "step": 96790
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2665316008347676e-05,
+      "loss": 0.938,
+      "step": 96800
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266455435891968e-05,
+      "loss": 0.9377,
+      "step": 96810
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266379270949168e-05,
+      "loss": 1.0049,
+      "step": 96820
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2663031060063675e-05,
+      "loss": 0.9461,
+      "step": 96830
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266226941063567e-05,
+      "loss": 0.9837,
+      "step": 96840
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266150776120768e-05,
+      "loss": 1.0025,
+      "step": 96850
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.266074611177967e-05,
+      "loss": 1.0845,
+      "step": 96860
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.265998446235167e-05,
+      "loss": 0.9352,
+      "step": 96870
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.265922281292367e-05,
+      "loss": 1.0126,
+      "step": 96880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2658461163495666e-05,
+      "loss": 0.9416,
+      "step": 96890
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.2657699514067664e-05,
+      "loss": 0.9623,
+      "step": 96900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265693786463967e-05,
+      "loss": 1.0735,
+      "step": 96910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.2656176215211666e-05,
+      "loss": 1.0162,
+      "step": 96920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265541456578366e-05,
+      "loss": 0.9823,
+      "step": 96930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265465291635567e-05,
+      "loss": 1.0118,
+      "step": 96940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265389126692766e-05,
+      "loss": 1.0616,
+      "step": 96950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265312961749966e-05,
+      "loss": 0.9281,
+      "step": 96960
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265236796807166e-05,
+      "loss": 0.9579,
+      "step": 96970
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265160631864366e-05,
+      "loss": 1.0405,
+      "step": 96980
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.2650844669215655e-05,
+      "loss": 0.9221,
+      "step": 96990
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.265008301978766e-05,
+      "loss": 1.0151,
+      "step": 97000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264932137035965e-05,
+      "loss": 0.95,
+      "step": 97010
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.2648559720931654e-05,
+      "loss": 1.0119,
+      "step": 97020
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264779807150365e-05,
+      "loss": 0.9632,
+      "step": 97030
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264703642207565e-05,
+      "loss": 1.019,
+      "step": 97040
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.2646274772647646e-05,
+      "loss": 1.0369,
+      "step": 97050
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264551312321965e-05,
+      "loss": 1.1966,
+      "step": 97060
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264475147379164e-05,
+      "loss": 0.9536,
+      "step": 97070
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.2643989824363645e-05,
+      "loss": 1.1326,
+      "step": 97080
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264322817493564e-05,
+      "loss": 1.0661,
+      "step": 97090
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264246652550764e-05,
+      "loss": 1.0547,
+      "step": 97100
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.264170487607964e-05,
+      "loss": 1.054,
+      "step": 97110
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.2640943226651635e-05,
+      "loss": 0.9641,
+      "step": 97120
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.264018157722364e-05,
+      "loss": 1.0248,
+      "step": 97130
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263941992779564e-05,
+      "loss": 1.0613,
+      "step": 97140
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2638658278367635e-05,
+      "loss": 0.9506,
+      "step": 97150
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263789662893963e-05,
+      "loss": 1.0234,
+      "step": 97160
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2637134979511636e-05,
+      "loss": 0.9712,
+      "step": 97170
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263637333008363e-05,
+      "loss": 1.025,
+      "step": 97180
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263561168065563e-05,
+      "loss": 0.9812,
+      "step": 97190
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263485003122763e-05,
+      "loss": 0.9144,
+      "step": 97200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2634088381799626e-05,
+      "loss": 0.9915,
+      "step": 97210
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2633326732371624e-05,
+      "loss": 0.9186,
+      "step": 97220
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263256508294363e-05,
+      "loss": 0.9985,
+      "step": 97230
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263180343351562e-05,
+      "loss": 0.9781,
+      "step": 97240
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263104178408762e-05,
+      "loss": 1.0269,
+      "step": 97250
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.263028013465962e-05,
+      "loss": 1.0579,
+      "step": 97260
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.262951848523162e-05,
+      "loss": 1.0474,
+      "step": 97270
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2628756835803615e-05,
+      "loss": 0.9836,
+      "step": 97280
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.262799518637562e-05,
+      "loss": 1.0005,
+      "step": 97290
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.262723353694762e-05,
+      "loss": 1.0837,
+      "step": 97300
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2626471887519614e-05,
+      "loss": 0.9685,
+      "step": 97310
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.262571023809161e-05,
+      "loss": 1.0015,
+      "step": 97320
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.262494858866361e-05,
+      "loss": 1.0488,
+      "step": 97330
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.2624186939235613e-05,
+      "loss": 1.0675,
+      "step": 97340
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.262342528980761e-05,
+      "loss": 0.9229,
+      "step": 97350
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.262266364037961e-05,
+      "loss": 0.9788,
+      "step": 97360
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2621901990951606e-05,
+      "loss": 1.1121,
+      "step": 97370
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.262114034152361e-05,
+      "loss": 0.9039,
+      "step": 97380
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.26203786920956e-05,
+      "loss": 1.0214,
+      "step": 97390
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2619617042667605e-05,
+      "loss": 1.0066,
+      "step": 97400
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.26188553932396e-05,
+      "loss": 1.0961,
+      "step": 97410
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.26180937438116e-05,
+      "loss": 0.9326,
+      "step": 97420
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.26173320943836e-05,
+      "loss": 0.9779,
+      "step": 97430
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.26165704449556e-05,
+      "loss": 1.0166,
+      "step": 97440
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.261580879552759e-05,
+      "loss": 1.0375,
+      "step": 97450
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.26150471460996e-05,
+      "loss": 0.9426,
+      "step": 97460
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2614285496671594e-05,
+      "loss": 0.8975,
+      "step": 97470
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.261352384724359e-05,
+      "loss": 1.0841,
+      "step": 97480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.261276219781559e-05,
+      "loss": 1.0511,
+      "step": 97490
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.261200054838759e-05,
+      "loss": 0.9427,
+      "step": 97500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.261123889895959e-05,
+      "loss": 1.0707,
+      "step": 97510
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.261047724953159e-05,
+      "loss": 1.015,
+      "step": 97520
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.260971560010359e-05,
+      "loss": 1.0752,
+      "step": 97530
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.260895395067558e-05,
+      "loss": 0.9221,
+      "step": 97540
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.260819230124759e-05,
+      "loss": 1.0106,
+      "step": 97550
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.260743065181958e-05,
+      "loss": 0.9072,
+      "step": 97560
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260666900239158e-05,
+      "loss": 0.9903,
+      "step": 97570
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260590735296358e-05,
+      "loss": 0.9152,
+      "step": 97580
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260514570353558e-05,
+      "loss": 0.9586,
+      "step": 97590
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2604384054107575e-05,
+      "loss": 0.9553,
+      "step": 97600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260362240467958e-05,
+      "loss": 0.9448,
+      "step": 97610
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260286075525157e-05,
+      "loss": 1.0008,
+      "step": 97620
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2602099105823574e-05,
+      "loss": 0.9353,
+      "step": 97630
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260133745639557e-05,
+      "loss": 1.0191,
+      "step": 97640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.260057580696757e-05,
+      "loss": 1.0302,
+      "step": 97650
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2599814157539566e-05,
+      "loss": 0.8996,
+      "step": 97660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259905250811157e-05,
+      "loss": 0.9337,
+      "step": 97670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259829085868357e-05,
+      "loss": 1.1112,
+      "step": 97680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2597529209255565e-05,
+      "loss": 1.0327,
+      "step": 97690
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259676755982756e-05,
+      "loss": 0.9229,
+      "step": 97700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259600591039956e-05,
+      "loss": 1.109,
+      "step": 97710
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2595244260971565e-05,
+      "loss": 0.9762,
+      "step": 97720
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259448261154356e-05,
+      "loss": 1.031,
+      "step": 97730
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259372096211556e-05,
+      "loss": 1.0787,
+      "step": 97740
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259295931268756e-05,
+      "loss": 0.999,
+      "step": 97750
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259219766325956e-05,
+      "loss": 1.0582,
+      "step": 97760
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.259143601383155e-05,
+      "loss": 1.0133,
+      "step": 97770
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2590674364403556e-05,
+      "loss": 1.0582,
+      "step": 97780
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2589912714975554e-05,
+      "loss": 1.1494,
+      "step": 97790
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258915106554755e-05,
+      "loss": 0.9635,
+      "step": 97800
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258838941611955e-05,
+      "loss": 0.9909,
+      "step": 97810
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258762776669155e-05,
+      "loss": 1.0182,
+      "step": 97820
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2586866117263544e-05,
+      "loss": 1.0336,
+      "step": 97830
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258610446783555e-05,
+      "loss": 0.9566,
+      "step": 97840
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2585342818407545e-05,
+      "loss": 0.9234,
+      "step": 97850
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258458116897954e-05,
+      "loss": 1.0787,
+      "step": 97860
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258381951955154e-05,
+      "loss": 1.0189,
+      "step": 97870
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2583057870123544e-05,
+      "loss": 0.9322,
+      "step": 97880
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258229622069554e-05,
+      "loss": 1.0308,
+      "step": 97890
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258153457126754e-05,
+      "loss": 1.1519,
+      "step": 97900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.258077292183954e-05,
+      "loss": 0.9651,
+      "step": 97910
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2580011272411534e-05,
+      "loss": 0.9792,
+      "step": 97920
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.257924962298354e-05,
+      "loss": 0.8744,
+      "step": 97930
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2578487973555536e-05,
+      "loss": 1.0064,
+      "step": 97940
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2577726324127533e-05,
+      "loss": 1.0183,
+      "step": 97950
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.257696467469953e-05,
+      "loss": 0.9234,
+      "step": 97960
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2576203025271535e-05,
+      "loss": 0.9668,
+      "step": 97970
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.2575441375843526e-05,
+      "loss": 1.0153,
+      "step": 97980
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.257467972641553e-05,
+      "loss": 1.0363,
+      "step": 97990
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.257391807698753e-05,
+      "loss": 0.9223,
+      "step": 98000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.2573156427559525e-05,
+      "loss": 1.0069,
+      "step": 98010
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.257239477813152e-05,
+      "loss": 1.0004,
+      "step": 98020
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.257163312870352e-05,
+      "loss": 1.0341,
+      "step": 98030
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.257087147927552e-05,
+      "loss": 1.0382,
+      "step": 98040
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.257010982984752e-05,
+      "loss": 1.0743,
+      "step": 98050
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256934818041952e-05,
+      "loss": 1.0653,
+      "step": 98060
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256858653099152e-05,
+      "loss": 0.9607,
+      "step": 98070
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.2567824881563514e-05,
+      "loss": 0.9632,
+      "step": 98080
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256706323213551e-05,
+      "loss": 1.0735,
+      "step": 98090
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.2566301582707516e-05,
+      "loss": 0.9182,
+      "step": 98100
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256553993327951e-05,
+      "loss": 0.9093,
+      "step": 98110
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256477828385151e-05,
+      "loss": 0.9401,
+      "step": 98120
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256401663442351e-05,
+      "loss": 1.01,
+      "step": 98130
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256325498499551e-05,
+      "loss": 1.0346,
+      "step": 98140
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.25624933355675e-05,
+      "loss": 1.0096,
+      "step": 98150
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.256173168613951e-05,
+      "loss": 1.0468,
+      "step": 98160
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.2560970036711505e-05,
+      "loss": 1.0752,
+      "step": 98170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.25602083872835e-05,
+      "loss": 1.049,
+      "step": 98180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.25594467378555e-05,
+      "loss": 0.9481,
+      "step": 98190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.2558685088427504e-05,
+      "loss": 0.9796,
+      "step": 98200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.2557923438999495e-05,
+      "loss": 1.0193,
+      "step": 98210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.25571617895715e-05,
+      "loss": 0.9794,
+      "step": 98220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2556400140143496e-05,
+      "loss": 1.1592,
+      "step": 98230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2555638490715494e-05,
+      "loss": 1.06,
+      "step": 98240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.255487684128749e-05,
+      "loss": 0.9621,
+      "step": 98250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2554115191859496e-05,
+      "loss": 1.191,
+      "step": 98260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.255335354243149e-05,
+      "loss": 1.0613,
+      "step": 98270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.255259189300349e-05,
+      "loss": 1.0551,
+      "step": 98280
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.255183024357549e-05,
+      "loss": 0.9556,
+      "step": 98290
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2551068594147485e-05,
+      "loss": 0.9229,
+      "step": 98300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.255030694471949e-05,
+      "loss": 1.0367,
+      "step": 98310
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254954529529149e-05,
+      "loss": 0.9341,
+      "step": 98320
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2548783645863485e-05,
+      "loss": 1.0384,
+      "step": 98330
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254802199643548e-05,
+      "loss": 0.8649,
+      "step": 98340
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2547260347007486e-05,
+      "loss": 1.0565,
+      "step": 98350
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254649869757948e-05,
+      "loss": 1.0396,
+      "step": 98360
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254573704815148e-05,
+      "loss": 0.97,
+      "step": 98370
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254497539872348e-05,
+      "loss": 1.0071,
+      "step": 98380
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2544213749295476e-05,
+      "loss": 0.9333,
+      "step": 98390
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.2543452099867474e-05,
+      "loss": 0.9812,
+      "step": 98400
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254269045043948e-05,
+      "loss": 1.1678,
+      "step": 98410
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254192880101147e-05,
+      "loss": 0.9689,
+      "step": 98420
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.254116715158347e-05,
+      "loss": 1.0344,
+      "step": 98430
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.254040550215547e-05,
+      "loss": 1.0276,
+      "step": 98440
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253964385272747e-05,
+      "loss": 0.9995,
+      "step": 98450
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2538882203299465e-05,
+      "loss": 1.0188,
+      "step": 98460
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253812055387146e-05,
+      "loss": 0.9746,
+      "step": 98470
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253735890444347e-05,
+      "loss": 0.9644,
+      "step": 98480
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2536597255015464e-05,
+      "loss": 1.0236,
+      "step": 98490
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253583560558746e-05,
+      "loss": 0.991,
+      "step": 98500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253507395615946e-05,
+      "loss": 1.0239,
+      "step": 98510
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2534312306731464e-05,
+      "loss": 0.9274,
+      "step": 98520
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2533550657303454e-05,
+      "loss": 1.1093,
+      "step": 98530
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253278900787546e-05,
+      "loss": 0.9726,
+      "step": 98540
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2532027358447456e-05,
+      "loss": 0.9669,
+      "step": 98550
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2531265709019454e-05,
+      "loss": 1.067,
+      "step": 98560
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.253050405959145e-05,
+      "loss": 1.0193,
+      "step": 98570
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2529742410163455e-05,
+      "loss": 0.9418,
+      "step": 98580
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2528980760735446e-05,
+      "loss": 0.9775,
+      "step": 98590
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.252821911130745e-05,
+      "loss": 0.9475,
+      "step": 98600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.252745746187945e-05,
+      "loss": 1.0438,
+      "step": 98610
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2526695812451445e-05,
+      "loss": 0.9041,
+      "step": 98620
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.252593416302344e-05,
+      "loss": 0.9707,
+      "step": 98630
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.252517251359545e-05,
+      "loss": 0.9503,
+      "step": 98640
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.2524410864167444e-05,
+      "loss": 1.0354,
+      "step": 98650
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.252364921473944e-05,
+      "loss": 0.9431,
+      "step": 98660
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.252288756531144e-05,
+      "loss": 1.0766,
+      "step": 98670
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.252212591588344e-05,
+      "loss": 0.9608,
+      "step": 98680
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.252136426645544e-05,
+      "loss": 1.0741,
+      "step": 98690
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.252060261702744e-05,
+      "loss": 1.0488,
+      "step": 98700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.2519840967599436e-05,
+      "loss": 0.9411,
+      "step": 98710
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251907931817143e-05,
+      "loss": 0.9875,
+      "step": 98720
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251831766874344e-05,
+      "loss": 0.9521,
+      "step": 98730
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251755601931543e-05,
+      "loss": 1.0612,
+      "step": 98740
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251679436988743e-05,
+      "loss": 0.9699,
+      "step": 98750
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251603272045943e-05,
+      "loss": 0.9858,
+      "step": 98760
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251527107103143e-05,
+      "loss": 1.0252,
+      "step": 98770
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.2514509421603425e-05,
+      "loss": 1.0965,
+      "step": 98780
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251374777217543e-05,
+      "loss": 0.9265,
+      "step": 98790
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251298612274742e-05,
+      "loss": 1.0139,
+      "step": 98800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.2512224473319424e-05,
+      "loss": 0.9411,
+      "step": 98810
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251146282389142e-05,
+      "loss": 1.0666,
+      "step": 98820
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.251070117446342e-05,
+      "loss": 0.9599,
+      "step": 98830
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.2509939525035416e-05,
+      "loss": 0.9316,
+      "step": 98840
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.250917787560742e-05,
+      "loss": 1.032,
+      "step": 98850
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.250841622617942e-05,
+      "loss": 1.0282,
+      "step": 98860
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.2507654576751416e-05,
+      "loss": 1.0677,
+      "step": 98870
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.250689292732341e-05,
+      "loss": 0.9794,
+      "step": 98880
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.250613127789541e-05,
+      "loss": 0.9201,
+      "step": 98890
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2505369628467415e-05,
+      "loss": 1.0599,
+      "step": 98900
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.250460797903941e-05,
+      "loss": 1.0185,
+      "step": 98910
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.250384632961141e-05,
+      "loss": 0.8775,
+      "step": 98920
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.250308468018341e-05,
+      "loss": 1.0765,
+      "step": 98930
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2502323030755405e-05,
+      "loss": 0.9478,
+      "step": 98940
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.25015613813274e-05,
+      "loss": 0.9691,
+      "step": 98950
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2500799731899406e-05,
+      "loss": 0.9292,
+      "step": 98960
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.25000380824714e-05,
+      "loss": 1.109,
+      "step": 98970
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.24992764330434e-05,
+      "loss": 0.9498,
+      "step": 98980
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.24985147836154e-05,
+      "loss": 1.0212,
+      "step": 98990
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2497753134187396e-05,
+      "loss": 1.0446,
+      "step": 99000
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2496991484759394e-05,
+      "loss": 1.0109,
+      "step": 99010
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.24962298353314e-05,
+      "loss": 1.0266,
+      "step": 99020
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2495468185903395e-05,
+      "loss": 1.0687,
+      "step": 99030
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.249470653647539e-05,
+      "loss": 1.1183,
+      "step": 99040
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.249394488704739e-05,
+      "loss": 0.9629,
+      "step": 99050
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.249318323761939e-05,
+      "loss": 0.8597,
+      "step": 99060
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.249242158819139e-05,
+      "loss": 0.9384,
+      "step": 99070
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.249165993876339e-05,
+      "loss": 1.0556,
+      "step": 99080
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.249089828933539e-05,
+      "loss": 1.1154,
+      "step": 99090
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2490136639907384e-05,
+      "loss": 1.0153,
+      "step": 99100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248937499047939e-05,
+      "loss": 1.0609,
+      "step": 99110
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248861334105138e-05,
+      "loss": 1.064,
+      "step": 99120
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2487851691623384e-05,
+      "loss": 0.9507,
+      "step": 99130
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248709004219538e-05,
+      "loss": 0.9572,
+      "step": 99140
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248632839276738e-05,
+      "loss": 1.1731,
+      "step": 99150
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2485566743339376e-05,
+      "loss": 1.0478,
+      "step": 99160
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248480509391138e-05,
+      "loss": 1.0841,
+      "step": 99170
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248404344448337e-05,
+      "loss": 1.0732,
+      "step": 99180
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2483281795055375e-05,
+      "loss": 0.9224,
+      "step": 99190
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248252014562737e-05,
+      "loss": 1.0161,
+      "step": 99200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248175849619937e-05,
+      "loss": 1.1133,
+      "step": 99210
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248099684677137e-05,
+      "loss": 1.0059,
+      "step": 99220
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.248023519734337e-05,
+      "loss": 1.0435,
+      "step": 99230
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.247947354791537e-05,
+      "loss": 0.9938,
+      "step": 99240
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.247871189848737e-05,
+      "loss": 1.0656,
+      "step": 99250
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2477950249059364e-05,
+      "loss": 0.9466,
+      "step": 99260
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.247718859963136e-05,
+      "loss": 1.0576,
+      "step": 99270
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2476426950203366e-05,
+      "loss": 1.0048,
+      "step": 99280
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2475665300775363e-05,
+      "loss": 0.9277,
+      "step": 99290
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.247490365134736e-05,
+      "loss": 1.0446,
+      "step": 99300
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.247414200191936e-05,
+      "loss": 0.9428,
+      "step": 99310
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.247338035249136e-05,
+      "loss": 1.0168,
+      "step": 99320
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.247261870306335e-05,
+      "loss": 1.011,
+      "step": 99330
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.247185705363536e-05,
+      "loss": 1.0068,
+      "step": 99340
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2471095404207355e-05,
+      "loss": 1.001,
+      "step": 99350
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.247033375477935e-05,
+      "loss": 1.0617,
+      "step": 99360
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246957210535135e-05,
+      "loss": 0.987,
+      "step": 99370
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2468810455923354e-05,
+      "loss": 0.9139,
+      "step": 99380
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2468048806495345e-05,
+      "loss": 1.1606,
+      "step": 99390
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246728715706735e-05,
+      "loss": 1.0101,
+      "step": 99400
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246652550763934e-05,
+      "loss": 0.9963,
+      "step": 99410
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2465763858211344e-05,
+      "loss": 0.8877,
+      "step": 99420
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246500220878334e-05,
+      "loss": 0.9211,
+      "step": 99430
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246424055935534e-05,
+      "loss": 1.0957,
+      "step": 99440
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246347890992734e-05,
+      "loss": 1.0042,
+      "step": 99450
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246271726049934e-05,
+      "loss": 1.1068,
+      "step": 99460
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246195561107134e-05,
+      "loss": 1.0464,
+      "step": 99470
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2461193961643336e-05,
+      "loss": 0.933,
+      "step": 99480
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.246043231221534e-05,
+      "loss": 1.12,
+      "step": 99490
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.245967066278733e-05,
+      "loss": 1.0528,
+      "step": 99500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2458909013359335e-05,
+      "loss": 1.0271,
+      "step": 99510
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.245814736393133e-05,
+      "loss": 0.9928,
+      "step": 99520
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.245738571450333e-05,
+      "loss": 1.0206,
+      "step": 99530
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245662406507533e-05,
+      "loss": 0.9951,
+      "step": 99540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245586241564733e-05,
+      "loss": 0.9612,
+      "step": 99550
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245510076621932e-05,
+      "loss": 1.0734,
+      "step": 99560
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2454339116791326e-05,
+      "loss": 1.1213,
+      "step": 99570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2453577467363324e-05,
+      "loss": 0.8519,
+      "step": 99580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245281581793532e-05,
+      "loss": 1.0249,
+      "step": 99590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245205416850732e-05,
+      "loss": 1.0214,
+      "step": 99600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.245129251907932e-05,
+      "loss": 1.0225,
+      "step": 99610
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2450530869651314e-05,
+      "loss": 0.9408,
+      "step": 99620
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244976922022332e-05,
+      "loss": 0.9579,
+      "step": 99630
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2449007570795315e-05,
+      "loss": 0.9632,
+      "step": 99640
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244824592136731e-05,
+      "loss": 1.0114,
+      "step": 99650
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244748427193932e-05,
+      "loss": 1.0272,
+      "step": 99660
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2446722622511315e-05,
+      "loss": 1.0522,
+      "step": 99670
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244596097308331e-05,
+      "loss": 0.9696,
+      "step": 99680
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244519932365531e-05,
+      "loss": 1.0569,
+      "step": 99690
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2444437674227314e-05,
+      "loss": 1.0566,
+      "step": 99700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2443676024799304e-05,
+      "loss": 0.9429,
+      "step": 99710
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244291437537131e-05,
+      "loss": 1.0873,
+      "step": 99720
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2442152725943306e-05,
+      "loss": 1.0487,
+      "step": 99730
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2441391076515304e-05,
+      "loss": 0.8594,
+      "step": 99740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.24406294270873e-05,
+      "loss": 0.9359,
+      "step": 99750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2439867777659305e-05,
+      "loss": 0.9666,
+      "step": 99760
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2439106128231296e-05,
+      "loss": 0.9684,
+      "step": 99770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.24383444788033e-05,
+      "loss": 1.0667,
+      "step": 99780
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.24375828293753e-05,
+      "loss": 0.9678,
+      "step": 99790
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2436821179947295e-05,
+      "loss": 1.0662,
+      "step": 99800
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.243605953051929e-05,
+      "loss": 1.0155,
+      "step": 99810
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.24352978810913e-05,
+      "loss": 0.9789,
+      "step": 99820
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2434536231663294e-05,
+      "loss": 1.0102,
+      "step": 99830
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.243377458223529e-05,
+      "loss": 1.1627,
+      "step": 99840
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.243301293280729e-05,
+      "loss": 1.0526,
+      "step": 99850
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.243225128337929e-05,
+      "loss": 1.0165,
+      "step": 99860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.243148963395129e-05,
+      "loss": 0.9735,
+      "step": 99870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.243072798452328e-05,
+      "loss": 0.9357,
+      "step": 99880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2429966335095286e-05,
+      "loss": 1.0689,
+      "step": 99890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2429204685667283e-05,
+      "loss": 0.9398,
+      "step": 99900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.242844303623928e-05,
+      "loss": 1.0142,
+      "step": 99910
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.242768138681128e-05,
+      "loss": 0.9275,
+      "step": 99920
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.242691973738328e-05,
+      "loss": 0.9272,
+      "step": 99930
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.242615808795527e-05,
+      "loss": 1.0384,
+      "step": 99940
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.242539643852728e-05,
+      "loss": 1.043,
+      "step": 99950
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.2424634789099275e-05,
+      "loss": 0.8517,
+      "step": 99960
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.242387313967127e-05,
+      "loss": 0.9833,
+      "step": 99970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.242311149024327e-05,
+      "loss": 1.0518,
+      "step": 99980
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2422349840815274e-05,
+      "loss": 0.9231,
+      "step": 99990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2421588191387265e-05,
+      "loss": 1.0793,
+      "step": 100000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.242082654195927e-05,
+      "loss": 1.0116,
+      "step": 100010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.242006489253127e-05,
+      "loss": 0.9282,
+      "step": 100020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2419303243103264e-05,
+      "loss": 0.9403,
+      "step": 100030
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241854159367527e-05,
+      "loss": 1.0317,
+      "step": 100040
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2417779944247266e-05,
+      "loss": 0.9495,
+      "step": 100050
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241701829481926e-05,
+      "loss": 0.9814,
+      "step": 100060
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241625664539126e-05,
+      "loss": 0.9639,
+      "step": 100070
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2415494995963265e-05,
+      "loss": 1.0214,
+      "step": 100080
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2414733346535256e-05,
+      "loss": 1.0938,
+      "step": 100090
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241397169710726e-05,
+      "loss": 0.9056,
+      "step": 100100
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241321004767926e-05,
+      "loss": 1.095,
+      "step": 100110
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2412448398251255e-05,
+      "loss": 0.941,
+      "step": 100120
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241168674882325e-05,
+      "loss": 1.0947,
+      "step": 100130
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2410925099395257e-05,
+      "loss": 0.9188,
+      "step": 100140
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.241016344996725e-05,
+      "loss": 0.9358,
+      "step": 100150
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.240940180053925e-05,
+      "loss": 1.0362,
+      "step": 100160
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.240864015111125e-05,
+      "loss": 0.8748,
+      "step": 100170
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2407878501683246e-05,
+      "loss": 1.0845,
+      "step": 100180
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2407116852255244e-05,
+      "loss": 1.0104,
+      "step": 100190
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240635520282725e-05,
+      "loss": 0.9225,
+      "step": 100200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240559355339924e-05,
+      "loss": 1.1116,
+      "step": 100210
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240483190397124e-05,
+      "loss": 1.1851,
+      "step": 100220
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240407025454324e-05,
+      "loss": 0.9595,
+      "step": 100230
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240330860511524e-05,
+      "loss": 1.0091,
+      "step": 100240
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240254695568724e-05,
+      "loss": 0.9862,
+      "step": 100250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240178530625924e-05,
+      "loss": 0.9961,
+      "step": 100260
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.240102365683124e-05,
+      "loss": 0.9902,
+      "step": 100270
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2400262007403235e-05,
+      "loss": 0.9856,
+      "step": 100280
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.239950035797524e-05,
+      "loss": 0.97,
+      "step": 100290
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.239873870854723e-05,
+      "loss": 1.0309,
+      "step": 100300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2397977059119234e-05,
+      "loss": 0.9651,
+      "step": 100310
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2397215409691224e-05,
+      "loss": 1.0051,
+      "step": 100320
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.239645376026323e-05,
+      "loss": 1.0564,
+      "step": 100330
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2395692110835226e-05,
+      "loss": 0.9442,
+      "step": 100340
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2394930461407224e-05,
+      "loss": 1.0412,
+      "step": 100350
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.239416881197922e-05,
+      "loss": 0.9583,
+      "step": 100360
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2393407162551225e-05,
+      "loss": 1.0199,
+      "step": 100370
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2392645513123216e-05,
+      "loss": 0.9451,
+      "step": 100380
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.239188386369522e-05,
+      "loss": 0.939,
+      "step": 100390
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.239112221426722e-05,
+      "loss": 1.0622,
+      "step": 100400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2390360564839215e-05,
+      "loss": 1.0828,
+      "step": 100410
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238959891541121e-05,
+      "loss": 0.9799,
+      "step": 100420
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238883726598322e-05,
+      "loss": 0.915,
+      "step": 100430
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2388075616555214e-05,
+      "loss": 1.0746,
+      "step": 100440
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238731396712721e-05,
+      "loss": 0.9926,
+      "step": 100450
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2386552317699216e-05,
+      "loss": 1.0119,
+      "step": 100460
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238579066827121e-05,
+      "loss": 1.0675,
+      "step": 100470
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238502901884321e-05,
+      "loss": 1.1353,
+      "step": 100480
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238426736941521e-05,
+      "loss": 1.043,
+      "step": 100490
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2383505719987206e-05,
+      "loss": 1.0472,
+      "step": 100500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2382744070559203e-05,
+      "loss": 0.8905,
+      "step": 100510
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.238198242113121e-05,
+      "loss": 1.0282,
+      "step": 100520
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.23812207717032e-05,
+      "loss": 1.0394,
+      "step": 100530
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.23804591222752e-05,
+      "loss": 1.0989,
+      "step": 100540
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.23796974728472e-05,
+      "loss": 1.0067,
+      "step": 100550
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.23789358234192e-05,
+      "loss": 1.0792,
+      "step": 100560
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2378174173991195e-05,
+      "loss": 0.972,
+      "step": 100570
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.23774125245632e-05,
+      "loss": 1.1283,
+      "step": 100580
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.237665087513519e-05,
+      "loss": 0.9603,
+      "step": 100590
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2375889225707194e-05,
+      "loss": 0.9506,
+      "step": 100600
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.237512757627919e-05,
+      "loss": 0.9857,
+      "step": 100610
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.237436592685119e-05,
+      "loss": 0.8924,
+      "step": 100620
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2373604277423193e-05,
+      "loss": 0.9821,
+      "step": 100630
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.237284262799519e-05,
+      "loss": 0.9867,
+      "step": 100640
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.237208097856719e-05,
+      "loss": 0.9118,
+      "step": 100650
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2371319329139186e-05,
+      "loss": 0.9618,
+      "step": 100660
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.237055767971119e-05,
+      "loss": 0.973,
+      "step": 100670
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236979603028318e-05,
+      "loss": 1.0299,
+      "step": 100680
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2369034380855185e-05,
+      "loss": 0.8698,
+      "step": 100690
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236827273142718e-05,
+      "loss": 1.0033,
+      "step": 100700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236751108199918e-05,
+      "loss": 1.0166,
+      "step": 100710
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236674943257118e-05,
+      "loss": 1.031,
+      "step": 100720
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236598778314318e-05,
+      "loss": 1.0282,
+      "step": 100730
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236522613371517e-05,
+      "loss": 1.0678,
+      "step": 100740
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2364464484287177e-05,
+      "loss": 1.0159,
+      "step": 100750
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2363702834859174e-05,
+      "loss": 1.0332,
+      "step": 100760
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236294118543117e-05,
+      "loss": 1.035,
+      "step": 100770
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.236217953600317e-05,
+      "loss": 1.0426,
+      "step": 100780
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2361417886575166e-05,
+      "loss": 0.9645,
+      "step": 100790
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2360656237147164e-05,
+      "loss": 0.9957,
+      "step": 100800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.235989458771917e-05,
+      "loss": 1.0889,
+      "step": 100810
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2359132938291166e-05,
+      "loss": 1.0109,
+      "step": 100820
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.235837128886316e-05,
+      "loss": 0.9453,
+      "step": 100830
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.235760963943517e-05,
+      "loss": 1.0748,
+      "step": 100840
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235684799000716e-05,
+      "loss": 1.0741,
+      "step": 100850
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235608634057916e-05,
+      "loss": 1.0557,
+      "step": 100860
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235532469115116e-05,
+      "loss": 1.1179,
+      "step": 100870
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235456304172316e-05,
+      "loss": 1.1716,
+      "step": 100880
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.2353801392295155e-05,
+      "loss": 1.0247,
+      "step": 100890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235303974286716e-05,
+      "loss": 1.0637,
+      "step": 100900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235227809343915e-05,
+      "loss": 0.8804,
+      "step": 100910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.2351516444011154e-05,
+      "loss": 0.967,
+      "step": 100920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.235075479458315e-05,
+      "loss": 0.9848,
+      "step": 100930
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234999314515515e-05,
+      "loss": 0.9814,
+      "step": 100940
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.2349231495727146e-05,
+      "loss": 0.9376,
+      "step": 100950
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234846984629915e-05,
+      "loss": 1.0327,
+      "step": 100960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234770819687114e-05,
+      "loss": 0.9505,
+      "step": 100970
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.2346946547443145e-05,
+      "loss": 0.9275,
+      "step": 100980
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234618489801514e-05,
+      "loss": 1.1435,
+      "step": 100990
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234542324858714e-05,
+      "loss": 0.9607,
+      "step": 101000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234466159915914e-05,
+      "loss": 0.9615,
+      "step": 101010
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234389994973114e-05,
+      "loss": 0.9239,
+      "step": 101020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234313830030314e-05,
+      "loss": 1.0066,
+      "step": 101030
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234237665087514e-05,
+      "loss": 1.0023,
+      "step": 101040
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234161500144714e-05,
+      "loss": 0.9722,
+      "step": 101050
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.234085335201913e-05,
+      "loss": 1.0157,
+      "step": 101060
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2340091702591136e-05,
+      "loss": 1.0726,
+      "step": 101070
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2339330053163134e-05,
+      "loss": 0.9872,
+      "step": 101080
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233856840373513e-05,
+      "loss": 0.9742,
+      "step": 101090
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233780675430713e-05,
+      "loss": 1.032,
+      "step": 101100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233704510487913e-05,
+      "loss": 1.0421,
+      "step": 101110
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2336283455451123e-05,
+      "loss": 1.0276,
+      "step": 101120
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233552180602313e-05,
+      "loss": 0.9651,
+      "step": 101130
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2334760156595125e-05,
+      "loss": 0.9837,
+      "step": 101140
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233399850716712e-05,
+      "loss": 0.9514,
+      "step": 101150
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233323685773912e-05,
+      "loss": 1.0041,
+      "step": 101160
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2332475208311124e-05,
+      "loss": 1.0001,
+      "step": 101170
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2331713558883115e-05,
+      "loss": 1.0262,
+      "step": 101180
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233095190945512e-05,
+      "loss": 1.0007,
+      "step": 101190
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.233019026002712e-05,
+      "loss": 1.0188,
+      "step": 101200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2329428610599114e-05,
+      "loss": 0.9765,
+      "step": 101210
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.232866696117111e-05,
+      "loss": 1.0574,
+      "step": 101220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.232790531174311e-05,
+      "loss": 0.9181,
+      "step": 101230
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2327143662315113e-05,
+      "loss": 0.9375,
+      "step": 101240
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.232638201288711e-05,
+      "loss": 1.0516,
+      "step": 101250
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.232562036345911e-05,
+      "loss": 1.0632,
+      "step": 101260
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2324858714031106e-05,
+      "loss": 1.0274,
+      "step": 101270
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.232409706460311e-05,
+      "loss": 0.9852,
+      "step": 101280
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.23233354151751e-05,
+      "loss": 1.0384,
+      "step": 101290
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2322573765747105e-05,
+      "loss": 1.0603,
+      "step": 101300
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.23218121163191e-05,
+      "loss": 1.0125,
+      "step": 101310
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.23210504668911e-05,
+      "loss": 1.0624,
+      "step": 101320
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.23202888174631e-05,
+      "loss": 1.1083,
+      "step": 101330
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.23195271680351e-05,
+      "loss": 1.0395,
+      "step": 101340
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231876551860709e-05,
+      "loss": 1.0247,
+      "step": 101350
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2318003869179097e-05,
+      "loss": 0.8795,
+      "step": 101360
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2317242219751094e-05,
+      "loss": 0.8692,
+      "step": 101370
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231648057032309e-05,
+      "loss": 0.9097,
+      "step": 101380
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231571892089509e-05,
+      "loss": 0.9196,
+      "step": 101390
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231495727146709e-05,
+      "loss": 0.9629,
+      "step": 101400
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231419562203909e-05,
+      "loss": 1.0664,
+      "step": 101410
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231343397261109e-05,
+      "loss": 0.9658,
+      "step": 101420
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231267232318309e-05,
+      "loss": 1.0317,
+      "step": 101430
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231191067375508e-05,
+      "loss": 0.8877,
+      "step": 101440
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.231114902432709e-05,
+      "loss": 0.9852,
+      "step": 101450
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2310387374899085e-05,
+      "loss": 1.0025,
+      "step": 101460
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.230962572547108e-05,
+      "loss": 0.9564,
+      "step": 101470
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.230886407604308e-05,
+      "loss": 0.8782,
+      "step": 101480
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2308102426615084e-05,
+      "loss": 1.0056,
+      "step": 101490
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2307340777187075e-05,
+      "loss": 1.1682,
+      "step": 101500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.230657912775908e-05,
+      "loss": 0.9702,
+      "step": 101510
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2305817478331076e-05,
+      "loss": 0.9066,
+      "step": 101520
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2305055828903074e-05,
+      "loss": 0.9796,
+      "step": 101530
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.230429417947507e-05,
+      "loss": 1.0806,
+      "step": 101540
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2303532530047076e-05,
+      "loss": 1.0416,
+      "step": 101550
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2302770880619066e-05,
+      "loss": 0.9953,
+      "step": 101560
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.230200923119107e-05,
+      "loss": 1.059,
+      "step": 101570
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.230124758176307e-05,
+      "loss": 1.0072,
+      "step": 101580
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2300485932335065e-05,
+      "loss": 1.0337,
+      "step": 101590
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229972428290706e-05,
+      "loss": 0.956,
+      "step": 101600
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229896263347907e-05,
+      "loss": 1.029,
+      "step": 101610
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2298200984051065e-05,
+      "loss": 1.0567,
+      "step": 101620
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229743933462306e-05,
+      "loss": 0.9947,
+      "step": 101630
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2296677685195066e-05,
+      "loss": 0.9803,
+      "step": 101640
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229591603576706e-05,
+      "loss": 0.9923,
+      "step": 101650
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229515438633906e-05,
+      "loss": 0.8974,
+      "step": 101660
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229439273691106e-05,
+      "loss": 0.9302,
+      "step": 101670
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2293631087483056e-05,
+      "loss": 0.9513,
+      "step": 101680
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2292869438055054e-05,
+      "loss": 1.0927,
+      "step": 101690
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229210778862705e-05,
+      "loss": 1.0158,
+      "step": 101700
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229134613919905e-05,
+      "loss": 0.9478,
+      "step": 101710
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.229058448977105e-05,
+      "loss": 1.0783,
+      "step": 101720
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2289822840343043e-05,
+      "loss": 1.0659,
+      "step": 101730
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228906119091505e-05,
+      "loss": 1.016,
+      "step": 101740
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2288299541487045e-05,
+      "loss": 1.0499,
+      "step": 101750
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228753789205904e-05,
+      "loss": 1.0607,
+      "step": 101760
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228677624263104e-05,
+      "loss": 1.0301,
+      "step": 101770
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2286014593203044e-05,
+      "loss": 0.9787,
+      "step": 101780
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228525294377504e-05,
+      "loss": 0.9644,
+      "step": 101790
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228449129434704e-05,
+      "loss": 1.0587,
+      "step": 101800
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228372964491904e-05,
+      "loss": 1.0416,
+      "step": 101810
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2282967995491034e-05,
+      "loss": 0.8108,
+      "step": 101820
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.228220634606304e-05,
+      "loss": 1.0105,
+      "step": 101830
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2281444696635036e-05,
+      "loss": 0.9056,
+      "step": 101840
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2280683047207033e-05,
+      "loss": 0.9325,
+      "step": 101850
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.227992139777903e-05,
+      "loss": 0.9782,
+      "step": 101860
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2279159748351035e-05,
+      "loss": 1.0003,
+      "step": 101870
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2278398098923026e-05,
+      "loss": 1.1358,
+      "step": 101880
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.227763644949503e-05,
+      "loss": 0.9943,
+      "step": 101890
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.227687480006703e-05,
+      "loss": 0.8884,
+      "step": 101900
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2276113150639025e-05,
+      "loss": 0.9994,
+      "step": 101910
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.227535150121102e-05,
+      "loss": 1.0337,
+      "step": 101920
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.227458985178303e-05,
+      "loss": 0.9778,
+      "step": 101930
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.227382820235502e-05,
+      "loss": 0.9861,
+      "step": 101940
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.227306655292702e-05,
+      "loss": 0.9026,
+      "step": 101950
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.227230490349902e-05,
+      "loss": 0.9593,
+      "step": 101960
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2271543254071017e-05,
+      "loss": 1.0344,
+      "step": 101970
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2270781604643014e-05,
+      "loss": 0.9647,
+      "step": 101980
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.227001995521502e-05,
+      "loss": 1.1225,
+      "step": 101990
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2269258305787016e-05,
+      "loss": 0.979,
+      "step": 102000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226849665635901e-05,
+      "loss": 1.0741,
+      "step": 102010
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226773500693101e-05,
+      "loss": 1.0312,
+      "step": 102020
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226697335750301e-05,
+      "loss": 0.9267,
+      "step": 102030
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226621170807501e-05,
+      "loss": 1.0939,
+      "step": 102040
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226545005864701e-05,
+      "loss": 1.0078,
+      "step": 102050
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226468840921901e-05,
+      "loss": 0.9081,
+      "step": 102060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2263926759791005e-05,
+      "loss": 1.037,
+      "step": 102070
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.226316511036301e-05,
+      "loss": 0.9424,
+      "step": 102080
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2262403460935e-05,
+      "loss": 0.9148,
+      "step": 102090
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2261641811507004e-05,
+      "loss": 1.0404,
+      "step": 102100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2260880162079e-05,
+      "loss": 0.9413,
+      "step": 102110
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2260118512651e-05,
+      "loss": 0.9919,
+      "step": 102120
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2259356863222996e-05,
+      "loss": 1.0158,
+      "step": 102130
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2258595213795e-05,
+      "loss": 1.0012,
+      "step": 102140
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.225783356436699e-05,
+      "loss": 1.003,
+      "step": 102150
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.2257071914938996e-05,
+      "loss": 0.9693,
+      "step": 102160
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.225631026551099e-05,
+      "loss": 1.0843,
+      "step": 102170
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.225554861608299e-05,
+      "loss": 1.0002,
+      "step": 102180
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.225478696665499e-05,
+      "loss": 1.0294,
+      "step": 102190
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.2254025317226985e-05,
+      "loss": 0.9447,
+      "step": 102200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.225326366779899e-05,
+      "loss": 0.9187,
+      "step": 102210
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.225250201837099e-05,
+      "loss": 1.1139,
+      "step": 102220
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.2251740368942985e-05,
+      "loss": 1.1534,
+      "step": 102230
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.225097871951498e-05,
+      "loss": 0.9732,
+      "step": 102240
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.2250217070086986e-05,
+      "loss": 1.0919,
+      "step": 102250
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224945542065898e-05,
+      "loss": 1.0054,
+      "step": 102260
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224869377123098e-05,
+      "loss": 0.9121,
+      "step": 102270
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224793212180298e-05,
+      "loss": 0.8787,
+      "step": 102280
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.2247170472374976e-05,
+      "loss": 1.1303,
+      "step": 102290
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.2246408822946974e-05,
+      "loss": 0.958,
+      "step": 102300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224564717351898e-05,
+      "loss": 1.0961,
+      "step": 102310
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224488552409097e-05,
+      "loss": 0.8847,
+      "step": 102320
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224412387466297e-05,
+      "loss": 0.9644,
+      "step": 102330
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224336222523497e-05,
+      "loss": 0.9715,
+      "step": 102340
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224260057580697e-05,
+      "loss": 1.0552,
+      "step": 102350
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.2241838926378965e-05,
+      "loss": 1.0424,
+      "step": 102360
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224107727695097e-05,
+      "loss": 1.0399,
+      "step": 102370
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.224031562752297e-05,
+      "loss": 0.889,
+      "step": 102380
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.2239553978094964e-05,
+      "loss": 0.8876,
+      "step": 102390
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223879232866696e-05,
+      "loss": 1.0067,
+      "step": 102400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223803067923896e-05,
+      "loss": 0.9554,
+      "step": 102410
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.2237269029810964e-05,
+      "loss": 0.9889,
+      "step": 102420
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223650738038296e-05,
+      "loss": 0.9881,
+      "step": 102430
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223574573095496e-05,
+      "loss": 0.9778,
+      "step": 102440
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.2234984081526956e-05,
+      "loss": 1.107,
+      "step": 102450
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223422243209896e-05,
+      "loss": 1.0785,
+      "step": 102460
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223346078267095e-05,
+      "loss": 0.8615,
+      "step": 102470
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.2232699133242955e-05,
+      "loss": 1.0174,
+      "step": 102480
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223193748381495e-05,
+      "loss": 1.0632,
+      "step": 102490
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223117583438695e-05,
+      "loss": 1.0121,
+      "step": 102500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.223041418495895e-05,
+      "loss": 1.0912,
+      "step": 102510
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222965253553095e-05,
+      "loss": 1.1862,
+      "step": 102520
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222889088610294e-05,
+      "loss": 1.0199,
+      "step": 102530
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222812923667495e-05,
+      "loss": 0.9465,
+      "step": 102540
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.2227367587246944e-05,
+      "loss": 0.9647,
+      "step": 102550
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222660593781894e-05,
+      "loss": 0.9968,
+      "step": 102560
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222584428839094e-05,
+      "loss": 1.0793,
+      "step": 102570
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222508263896294e-05,
+      "loss": 1.036,
+      "step": 102580
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222432098953494e-05,
+      "loss": 0.9787,
+      "step": 102590
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.222355934010694e-05,
+      "loss": 1.0532,
+      "step": 102600
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.2222797690678936e-05,
+      "loss": 1.0106,
+      "step": 102610
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.222203604125093e-05,
+      "loss": 1.0997,
+      "step": 102620
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.222127439182294e-05,
+      "loss": 1.0913,
+      "step": 102630
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.222051274239493e-05,
+      "loss": 1.0221,
+      "step": 102640
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221975109296693e-05,
+      "loss": 1.06,
+      "step": 102650
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221898944353893e-05,
+      "loss": 1.0369,
+      "step": 102660
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221822779411093e-05,
+      "loss": 1.0743,
+      "step": 102670
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.2217466144682925e-05,
+      "loss": 1.0004,
+      "step": 102680
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221670449525493e-05,
+      "loss": 0.9822,
+      "step": 102690
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221594284582692e-05,
+      "loss": 0.939,
+      "step": 102700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.2215181196398924e-05,
+      "loss": 0.9512,
+      "step": 102710
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221441954697092e-05,
+      "loss": 1.1222,
+      "step": 102720
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221365789754292e-05,
+      "loss": 0.9477,
+      "step": 102730
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.2212896248114916e-05,
+      "loss": 0.9842,
+      "step": 102740
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221213459868692e-05,
+      "loss": 0.9976,
+      "step": 102750
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.221137294925892e-05,
+      "loss": 0.9867,
+      "step": 102760
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.2210611299830916e-05,
+      "loss": 0.9802,
+      "step": 102770
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.220984965040291e-05,
+      "loss": 0.9317,
+      "step": 102780
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.220908800097491e-05,
+      "loss": 0.9555,
+      "step": 102790
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.2208326351546915e-05,
+      "loss": 0.94,
+      "step": 102800
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.220756470211891e-05,
+      "loss": 0.9063,
+      "step": 102810
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.220680305269091e-05,
+      "loss": 0.973,
+      "step": 102820
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.220604140326291e-05,
+      "loss": 1.0045,
+      "step": 102830
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.220527975383491e-05,
+      "loss": 0.9999,
+      "step": 102840
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.22045181044069e-05,
+      "loss": 0.965,
+      "step": 102850
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2203756454978906e-05,
+      "loss": 0.9443,
+      "step": 102860
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2202994805550904e-05,
+      "loss": 0.9717,
+      "step": 102870
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.22022331561229e-05,
+      "loss": 0.9516,
+      "step": 102880
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.22014715066949e-05,
+      "loss": 0.9303,
+      "step": 102890
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.22007098572669e-05,
+      "loss": 1.1099,
+      "step": 102900
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2199948207838894e-05,
+      "loss": 0.898,
+      "step": 102910
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.21991865584109e-05,
+      "loss": 0.9275,
+      "step": 102920
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2198424908982895e-05,
+      "loss": 1.083,
+      "step": 102930
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.219766325955489e-05,
+      "loss": 1.1317,
+      "step": 102940
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.219690161012689e-05,
+      "loss": 0.9893,
+      "step": 102950
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2196139960698895e-05,
+      "loss": 1.0601,
+      "step": 102960
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.219537831127089e-05,
+      "loss": 1.0426,
+      "step": 102970
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.219461666184289e-05,
+      "loss": 0.8648,
+      "step": 102980
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.219385501241489e-05,
+      "loss": 0.9606,
+      "step": 102990
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2193093362986884e-05,
+      "loss": 0.8779,
+      "step": 103000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.219233171355889e-05,
+      "loss": 0.9287,
+      "step": 103010
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2191570064130886e-05,
+      "loss": 1.0877,
+      "step": 103020
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.2190808414702884e-05,
+      "loss": 1.1372,
+      "step": 103030
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.219004676527488e-05,
+      "loss": 1.0965,
+      "step": 103040
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.2189285115846885e-05,
+      "loss": 0.9761,
+      "step": 103050
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.2188523466418876e-05,
+      "loss": 1.0043,
+      "step": 103060
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218776181699088e-05,
+      "loss": 0.9723,
+      "step": 103070
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218700016756287e-05,
+      "loss": 1.085,
+      "step": 103080
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.2186238518134875e-05,
+      "loss": 0.9555,
+      "step": 103090
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218547686870687e-05,
+      "loss": 1.043,
+      "step": 103100
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218471521927887e-05,
+      "loss": 0.9363,
+      "step": 103110
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218395356985087e-05,
+      "loss": 0.8972,
+      "step": 103120
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218319192042287e-05,
+      "loss": 1.081,
+      "step": 103130
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218243027099487e-05,
+      "loss": 1.0756,
+      "step": 103140
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218166862156687e-05,
+      "loss": 0.9799,
+      "step": 103150
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.2180906972138864e-05,
+      "loss": 0.9857,
+      "step": 103160
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.218014532271086e-05,
+      "loss": 0.9651,
+      "step": 103170
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.2179383673282866e-05,
+      "loss": 0.8486,
+      "step": 103180
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.217862202385486e-05,
+      "loss": 1.0492,
+      "step": 103190
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.217786037442686e-05,
+      "loss": 0.8882,
+      "step": 103200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.217709872499886e-05,
+      "loss": 1.0641,
+      "step": 103210
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.217633707557086e-05,
+      "loss": 1.0472,
+      "step": 103220
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.217557542614285e-05,
+      "loss": 0.9336,
+      "step": 103230
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.217481377671486e-05,
+      "loss": 0.984,
+      "step": 103240
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.2174052127286855e-05,
+      "loss": 1.139,
+      "step": 103250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.217329047785885e-05,
+      "loss": 1.0389,
+      "step": 103260
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.217252882843085e-05,
+      "loss": 0.9532,
+      "step": 103270
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2171767179002854e-05,
+      "loss": 1.0812,
+      "step": 103280
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2171005529574845e-05,
+      "loss": 0.9108,
+      "step": 103290
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.217024388014685e-05,
+      "loss": 0.865,
+      "step": 103300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2169482230718847e-05,
+      "loss": 0.9998,
+      "step": 103310
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2168720581290844e-05,
+      "loss": 1.0566,
+      "step": 103320
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216795893186284e-05,
+      "loss": 1.1027,
+      "step": 103330
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2167197282434846e-05,
+      "loss": 0.8877,
+      "step": 103340
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216643563300684e-05,
+      "loss": 1.0495,
+      "step": 103350
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216567398357884e-05,
+      "loss": 0.9926,
+      "step": 103360
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216491233415084e-05,
+      "loss": 0.8362,
+      "step": 103370
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2164150684722836e-05,
+      "loss": 1.0106,
+      "step": 103380
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216338903529484e-05,
+      "loss": 1.0732,
+      "step": 103390
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216262738586684e-05,
+      "loss": 0.9536,
+      "step": 103400
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2161865736438835e-05,
+      "loss": 0.9639,
+      "step": 103410
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.216110408701083e-05,
+      "loss": 0.9527,
+      "step": 103420
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2160342437582836e-05,
+      "loss": 1.1044,
+      "step": 103430
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.215958078815483e-05,
+      "loss": 0.9558,
+      "step": 103440
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.215881913872683e-05,
+      "loss": 0.9999,
+      "step": 103450
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.215805748929883e-05,
+      "loss": 0.9966,
+      "step": 103460
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2157295839870826e-05,
+      "loss": 0.979,
+      "step": 103470
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2156534190442824e-05,
+      "loss": 0.9265,
+      "step": 103480
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215577254101483e-05,
+      "loss": 0.8282,
+      "step": 103490
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215501089158682e-05,
+      "loss": 1.1051,
+      "step": 103500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215424924215882e-05,
+      "loss": 1.0701,
+      "step": 103510
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215348759273082e-05,
+      "loss": 1.1265,
+      "step": 103520
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215272594330282e-05,
+      "loss": 0.8984,
+      "step": 103530
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2151964293874815e-05,
+      "loss": 0.8973,
+      "step": 103540
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215120264444681e-05,
+      "loss": 1.0391,
+      "step": 103550
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.215044099501882e-05,
+      "loss": 1.1288,
+      "step": 103560
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2149679345590815e-05,
+      "loss": 0.9892,
+      "step": 103570
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.214891769616281e-05,
+      "loss": 0.9906,
+      "step": 103580
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.214815604673481e-05,
+      "loss": 1.0516,
+      "step": 103590
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2147394397306814e-05,
+      "loss": 1.039,
+      "step": 103600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2146632747878804e-05,
+      "loss": 0.9736,
+      "step": 103610
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.214587109845081e-05,
+      "loss": 0.9365,
+      "step": 103620
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2145109449022806e-05,
+      "loss": 1.0557,
+      "step": 103630
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2144347799594804e-05,
+      "loss": 1.0101,
+      "step": 103640
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.21435861501668e-05,
+      "loss": 0.898,
+      "step": 103650
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2142824500738805e-05,
+      "loss": 1.0131,
+      "step": 103660
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.2142062851310796e-05,
+      "loss": 1.0768,
+      "step": 103670
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.21413012018828e-05,
+      "loss": 0.848,
+      "step": 103680
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.21405395524548e-05,
+      "loss": 1.0353,
+      "step": 103690
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.2139777903026795e-05,
+      "loss": 0.9029,
+      "step": 103700
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213901625359879e-05,
+      "loss": 1.0748,
+      "step": 103710
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.21382546041708e-05,
+      "loss": 1.173,
+      "step": 103720
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.2137492954742794e-05,
+      "loss": 0.8621,
+      "step": 103730
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213673130531479e-05,
+      "loss": 1.0986,
+      "step": 103740
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213596965588679e-05,
+      "loss": 1.0211,
+      "step": 103750
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213520800645879e-05,
+      "loss": 1.0741,
+      "step": 103760
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213444635703079e-05,
+      "loss": 1.0372,
+      "step": 103770
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213368470760279e-05,
+      "loss": 0.9991,
+      "step": 103780
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.2132923058174786e-05,
+      "loss": 1.0695,
+      "step": 103790
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213216140874678e-05,
+      "loss": 0.996,
+      "step": 103800
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213139975931879e-05,
+      "loss": 1.0187,
+      "step": 103810
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.213063810989078e-05,
+      "loss": 1.0365,
+      "step": 103820
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212987646046278e-05,
+      "loss": 1.0668,
+      "step": 103830
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212911481103478e-05,
+      "loss": 1.1569,
+      "step": 103840
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212835316160678e-05,
+      "loss": 1.0461,
+      "step": 103850
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.2127591512178775e-05,
+      "loss": 0.9651,
+      "step": 103860
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212682986275078e-05,
+      "loss": 1.0441,
+      "step": 103870
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212606821332277e-05,
+      "loss": 0.9825,
+      "step": 103880
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.2125306563894774e-05,
+      "loss": 0.9342,
+      "step": 103890
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212454491446677e-05,
+      "loss": 0.9811,
+      "step": 103900
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.212378326503877e-05,
+      "loss": 1.0424,
+      "step": 103910
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2123021615610767e-05,
+      "loss": 0.9789,
+      "step": 103920
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.212225996618277e-05,
+      "loss": 0.9952,
+      "step": 103930
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.212149831675477e-05,
+      "loss": 0.9718,
+      "step": 103940
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2120736667326766e-05,
+      "loss": 0.9996,
+      "step": 103950
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211997501789876e-05,
+      "loss": 1.0053,
+      "step": 103960
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211921336847076e-05,
+      "loss": 0.9788,
+      "step": 103970
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2118451719042765e-05,
+      "loss": 1.1226,
+      "step": 103980
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211769006961476e-05,
+      "loss": 1.0317,
+      "step": 103990
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211692842018676e-05,
+      "loss": 1.05,
+      "step": 104000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211616677075876e-05,
+      "loss": 0.9724,
+      "step": 104010
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2115405121330755e-05,
+      "loss": 0.978,
+      "step": 104020
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211464347190275e-05,
+      "loss": 0.8924,
+      "step": 104030
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2113881822474756e-05,
+      "loss": 0.9322,
+      "step": 104040
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211312017304675e-05,
+      "loss": 0.9816,
+      "step": 104050
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211235852361875e-05,
+      "loss": 1.064,
+      "step": 104060
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.211159687419075e-05,
+      "loss": 0.9724,
+      "step": 104070
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2110835224762746e-05,
+      "loss": 0.9689,
+      "step": 104080
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2110073575334744e-05,
+      "loss": 1.0555,
+      "step": 104090
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.210931192590675e-05,
+      "loss": 1.0379,
+      "step": 104100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.210855027647874e-05,
+      "loss": 0.9693,
+      "step": 104110
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.210778862705074e-05,
+      "loss": 0.973,
+      "step": 104120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.210702697762274e-05,
+      "loss": 0.9458,
+      "step": 104130
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210626532819474e-05,
+      "loss": 0.9799,
+      "step": 104140
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210550367876674e-05,
+      "loss": 1.03,
+      "step": 104150
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210474202933874e-05,
+      "loss": 1.1929,
+      "step": 104160
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210398037991074e-05,
+      "loss": 0.9171,
+      "step": 104170
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.2103218730482735e-05,
+      "loss": 1.0877,
+      "step": 104180
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210245708105474e-05,
+      "loss": 1.0526,
+      "step": 104190
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210169543162673e-05,
+      "loss": 0.9059,
+      "step": 104200
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.2100933782198734e-05,
+      "loss": 0.9765,
+      "step": 104210
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.210017213277073e-05,
+      "loss": 1.0429,
+      "step": 104220
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209941048334273e-05,
+      "loss": 1.1204,
+      "step": 104230
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.2098648833914726e-05,
+      "loss": 0.9237,
+      "step": 104240
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209788718448673e-05,
+      "loss": 0.9751,
+      "step": 104250
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209712553505872e-05,
+      "loss": 1.1494,
+      "step": 104260
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.2096363885630725e-05,
+      "loss": 0.9362,
+      "step": 104270
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209560223620272e-05,
+      "loss": 0.983,
+      "step": 104280
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209484058677472e-05,
+      "loss": 0.9534,
+      "step": 104290
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209407893734672e-05,
+      "loss": 0.9235,
+      "step": 104300
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209331728791872e-05,
+      "loss": 1.0216,
+      "step": 104310
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209255563849072e-05,
+      "loss": 1.0057,
+      "step": 104320
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209179398906272e-05,
+      "loss": 0.9105,
+      "step": 104330
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.2091032339634714e-05,
+      "loss": 0.883,
+      "step": 104340
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.209027069020671e-05,
+      "loss": 0.9237,
+      "step": 104350
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2089509040778716e-05,
+      "loss": 1.1398,
+      "step": 104360
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2088747391350714e-05,
+      "loss": 1.0822,
+      "step": 104370
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.208798574192271e-05,
+      "loss": 0.8367,
+      "step": 104380
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.208722409249471e-05,
+      "loss": 1.0616,
+      "step": 104390
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.208646244306671e-05,
+      "loss": 0.9578,
+      "step": 104400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2085700793638703e-05,
+      "loss": 0.9024,
+      "step": 104410
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.208493914421071e-05,
+      "loss": 0.9695,
+      "step": 104420
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2084177494782705e-05,
+      "loss": 1.0767,
+      "step": 104430
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.20834158453547e-05,
+      "loss": 1.0597,
+      "step": 104440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.20826541959267e-05,
+      "loss": 0.9862,
+      "step": 104450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.20818925464987e-05,
+      "loss": 1.0477,
+      "step": 104460
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2081130897070695e-05,
+      "loss": 0.9006,
+      "step": 104470
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.20803692476427e-05,
+      "loss": 0.9066,
+      "step": 104480
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207960759821469e-05,
+      "loss": 0.8844,
+      "step": 104490
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2078845948786694e-05,
+      "loss": 1.0573,
+      "step": 104500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207808429935869e-05,
+      "loss": 1.0239,
+      "step": 104510
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207732264993069e-05,
+      "loss": 1.0198,
+      "step": 104520
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207656100050269e-05,
+      "loss": 0.9981,
+      "step": 104530
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207579935107469e-05,
+      "loss": 0.888,
+      "step": 104540
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207503770164669e-05,
+      "loss": 0.942,
+      "step": 104550
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.2074276052218686e-05,
+      "loss": 1.0587,
+      "step": 104560
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.207351440279069e-05,
+      "loss": 1.1142,
+      "step": 104570
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.207275275336268e-05,
+      "loss": 0.9867,
+      "step": 104580
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2071991103934685e-05,
+      "loss": 0.8811,
+      "step": 104590
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.207122945450668e-05,
+      "loss": 0.9559,
+      "step": 104600
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.207046780507868e-05,
+      "loss": 0.9567,
+      "step": 104610
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206970615565068e-05,
+      "loss": 1.2146,
+      "step": 104620
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206894450622268e-05,
+      "loss": 1.0032,
+      "step": 104630
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206818285679467e-05,
+      "loss": 1.2303,
+      "step": 104640
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2067421207366676e-05,
+      "loss": 1.1467,
+      "step": 104650
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2066659557938674e-05,
+      "loss": 1.0266,
+      "step": 104660
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206589790851067e-05,
+      "loss": 1.104,
+      "step": 104670
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206513625908267e-05,
+      "loss": 1.0796,
+      "step": 104680
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206437460965467e-05,
+      "loss": 0.988,
+      "step": 104690
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2063612960226664e-05,
+      "loss": 0.9162,
+      "step": 104700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206285131079867e-05,
+      "loss": 0.9854,
+      "step": 104710
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2062089661370666e-05,
+      "loss": 1.0281,
+      "step": 104720
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206132801194266e-05,
+      "loss": 0.9487,
+      "step": 104730
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.206056636251467e-05,
+      "loss": 1.0399,
+      "step": 104740
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2059804713086665e-05,
+      "loss": 1.0044,
+      "step": 104750
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.205904306365866e-05,
+      "loss": 1.0748,
+      "step": 104760
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.205828141423066e-05,
+      "loss": 1.1321,
+      "step": 104770
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2057519764802664e-05,
+      "loss": 1.102,
+      "step": 104780
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.2056758115374655e-05,
+      "loss": 0.9864,
+      "step": 104790
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.205599646594666e-05,
+      "loss": 0.9712,
+      "step": 104800
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2055234816518656e-05,
+      "loss": 0.9686,
+      "step": 104810
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2054473167090654e-05,
+      "loss": 0.8975,
+      "step": 104820
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.205371151766265e-05,
+      "loss": 1.0571,
+      "step": 104830
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2052949868234655e-05,
+      "loss": 1.016,
+      "step": 104840
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2052188218806646e-05,
+      "loss": 0.9893,
+      "step": 104850
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.205142656937865e-05,
+      "loss": 0.9589,
+      "step": 104860
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.205066491995065e-05,
+      "loss": 1.0901,
+      "step": 104870
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2049903270522645e-05,
+      "loss": 0.9488,
+      "step": 104880
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204914162109464e-05,
+      "loss": 1.0464,
+      "step": 104890
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204837997166665e-05,
+      "loss": 1.0126,
+      "step": 104900
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204761832223864e-05,
+      "loss": 0.9062,
+      "step": 104910
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204685667281064e-05,
+      "loss": 1.0518,
+      "step": 104920
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204609502338264e-05,
+      "loss": 0.922,
+      "step": 104930
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204533337395464e-05,
+      "loss": 1.02,
+      "step": 104940
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204457172452664e-05,
+      "loss": 1.0558,
+      "step": 104950
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204381007509863e-05,
+      "loss": 0.987,
+      "step": 104960
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2043048425670636e-05,
+      "loss": 0.982,
+      "step": 104970
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2042286776242634e-05,
+      "loss": 1.1043,
+      "step": 104980
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204152512681463e-05,
+      "loss": 0.9624,
+      "step": 104990
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.204076347738663e-05,
+      "loss": 0.9204,
+      "step": 105000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.204000182795863e-05,
+      "loss": 1.0293,
+      "step": 105010
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2039240178530623e-05,
+      "loss": 0.9007,
+      "step": 105020
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203847852910263e-05,
+      "loss": 0.9729,
+      "step": 105030
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2037716879674625e-05,
+      "loss": 0.8963,
+      "step": 105040
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203695523024662e-05,
+      "loss": 1.0172,
+      "step": 105050
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203619358081862e-05,
+      "loss": 0.9107,
+      "step": 105060
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2035431931390624e-05,
+      "loss": 0.9071,
+      "step": 105070
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2034670281962615e-05,
+      "loss": 0.9346,
+      "step": 105080
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203390863253462e-05,
+      "loss": 1.0253,
+      "step": 105090
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203314698310662e-05,
+      "loss": 0.9433,
+      "step": 105100
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2032385333678614e-05,
+      "loss": 1.0042,
+      "step": 105110
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203162368425062e-05,
+      "loss": 1.0543,
+      "step": 105120
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2030862034822616e-05,
+      "loss": 0.9946,
+      "step": 105130
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.203010038539461e-05,
+      "loss": 1.0103,
+      "step": 105140
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.202933873596661e-05,
+      "loss": 0.9499,
+      "step": 105150
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2028577086538615e-05,
+      "loss": 1.0518,
+      "step": 105160
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2027815437110606e-05,
+      "loss": 0.8697,
+      "step": 105170
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.202705378768261e-05,
+      "loss": 1.0938,
+      "step": 105180
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.202629213825461e-05,
+      "loss": 0.9393,
+      "step": 105190
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.2025530488826605e-05,
+      "loss": 1.1354,
+      "step": 105200
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.20247688393986e-05,
+      "loss": 0.9912,
+      "step": 105210
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.202400718997061e-05,
+      "loss": 0.8427,
+      "step": 105220
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.20232455405426e-05,
+      "loss": 0.9793,
+      "step": 105230
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.20224838911146e-05,
+      "loss": 1.0283,
+      "step": 105240
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.20217222416866e-05,
+      "loss": 0.8922,
+      "step": 105250
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2020960592258596e-05,
+      "loss": 1.0067,
+      "step": 105260
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2020198942830594e-05,
+      "loss": 1.0348,
+      "step": 105270
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.20194372934026e-05,
+      "loss": 0.9733,
+      "step": 105280
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201867564397459e-05,
+      "loss": 1.0114,
+      "step": 105290
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201791399454659e-05,
+      "loss": 0.9184,
+      "step": 105300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201715234511859e-05,
+      "loss": 1.0686,
+      "step": 105310
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201639069569059e-05,
+      "loss": 1.0006,
+      "step": 105320
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201562904626259e-05,
+      "loss": 0.9881,
+      "step": 105330
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201486739683459e-05,
+      "loss": 0.9944,
+      "step": 105340
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201410574740659e-05,
+      "loss": 0.8523,
+      "step": 105350
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2013344097978585e-05,
+      "loss": 1.1328,
+      "step": 105360
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201258244855059e-05,
+      "loss": 0.9941,
+      "step": 105370
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.201182079912258e-05,
+      "loss": 1.1374,
+      "step": 105380
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2011059149694584e-05,
+      "loss": 0.9337,
+      "step": 105390
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2010297500266575e-05,
+      "loss": 1.1266,
+      "step": 105400
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.200953585083858e-05,
+      "loss": 0.9654,
+      "step": 105410
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2008774201410576e-05,
+      "loss": 1.0181,
+      "step": 105420
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.2008012551982574e-05,
+      "loss": 1.0229,
+      "step": 105430
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.200725090255457e-05,
+      "loss": 1.0215,
+      "step": 105440
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.2006489253126575e-05,
+      "loss": 1.1407,
+      "step": 105450
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.2005727603698566e-05,
+      "loss": 1.0042,
+      "step": 105460
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.200496595427057e-05,
+      "loss": 0.8785,
+      "step": 105470
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.200420430484257e-05,
+      "loss": 0.9901,
+      "step": 105480
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.2003442655414565e-05,
+      "loss": 1.0999,
+      "step": 105490
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.200268100598656e-05,
+      "loss": 1.0493,
+      "step": 105500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.200191935655857e-05,
+      "loss": 0.8902,
+      "step": 105510
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.2001157707130565e-05,
+      "loss": 1.071,
+      "step": 105520
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.200039605770256e-05,
+      "loss": 1.0967,
+      "step": 105530
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.1999634408274566e-05,
+      "loss": 0.9589,
+      "step": 105540
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199887275884656e-05,
+      "loss": 0.9252,
+      "step": 105550
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199811110941856e-05,
+      "loss": 1.1385,
+      "step": 105560
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199734945999056e-05,
+      "loss": 0.9153,
+      "step": 105570
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.1996587810562556e-05,
+      "loss": 1.0352,
+      "step": 105580
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.1995826161134554e-05,
+      "loss": 0.9762,
+      "step": 105590
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199506451170656e-05,
+      "loss": 1.0185,
+      "step": 105600
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199430286227855e-05,
+      "loss": 1.2406,
+      "step": 105610
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199354121285055e-05,
+      "loss": 1.0934,
+      "step": 105620
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199277956342255e-05,
+      "loss": 0.9881,
+      "step": 105630
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199201791399455e-05,
+      "loss": 0.9934,
+      "step": 105640
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.1991256264566545e-05,
+      "loss": 0.9414,
+      "step": 105650
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.199049461513855e-05,
+      "loss": 0.892,
+      "step": 105660
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198973296571054e-05,
+      "loss": 1.0433,
+      "step": 105670
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.1988971316282544e-05,
+      "loss": 1.0145,
+      "step": 105680
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198820966685454e-05,
+      "loss": 0.9899,
+      "step": 105690
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198744801742654e-05,
+      "loss": 0.9218,
+      "step": 105700
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198668636799854e-05,
+      "loss": 0.9716,
+      "step": 105710
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198592471857054e-05,
+      "loss": 1.0256,
+      "step": 105720
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198516306914254e-05,
+      "loss": 1.0317,
+      "step": 105730
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.1984401419714536e-05,
+      "loss": 1.0023,
+      "step": 105740
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198363977028654e-05,
+      "loss": 0.9441,
+      "step": 105750
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198287812085853e-05,
+      "loss": 0.9836,
+      "step": 105760
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.1982116471430535e-05,
+      "loss": 0.9891,
+      "step": 105770
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198135482200253e-05,
+      "loss": 0.9221,
+      "step": 105780
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.198059317257453e-05,
+      "loss": 0.9407,
+      "step": 105790
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197983152314653e-05,
+      "loss": 0.9092,
+      "step": 105800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197906987371853e-05,
+      "loss": 0.9703,
+      "step": 105810
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197830822429052e-05,
+      "loss": 1.1683,
+      "step": 105820
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197754657486253e-05,
+      "loss": 0.8842,
+      "step": 105830
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197678492543452e-05,
+      "loss": 1.0915,
+      "step": 105840
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197602327600652e-05,
+      "loss": 0.9916,
+      "step": 105850
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.197526162657852e-05,
+      "loss": 1.0047,
+      "step": 105860
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.1974499977150517e-05,
+      "loss": 0.9969,
+      "step": 105870
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.1973738327722514e-05,
+      "loss": 1.0693,
+      "step": 105880
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.197297667829452e-05,
+      "loss": 1.0959,
+      "step": 105890
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.1972215028866516e-05,
+      "loss": 0.9761,
+      "step": 105900
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.197145337943851e-05,
+      "loss": 0.8377,
+      "step": 105910
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.197069173001052e-05,
+      "loss": 1.0948,
+      "step": 105920
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.196993008058251e-05,
+      "loss": 0.9084,
+      "step": 105930
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.196916843115451e-05,
+      "loss": 0.9184,
+      "step": 105940
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.196840678172651e-05,
+      "loss": 1.0557,
+      "step": 105950
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.196764513229851e-05,
+      "loss": 0.9935,
+      "step": 105960
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.1966883482870505e-05,
+      "loss": 1.0967,
+      "step": 105970
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.196612183344251e-05,
+      "loss": 0.954,
+      "step": 105980
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.19653601840145e-05,
+      "loss": 0.9556,
+      "step": 105990
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.1964598534586504e-05,
+      "loss": 0.9778,
+      "step": 106000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.19638368851585e-05,
+      "loss": 0.954,
+      "step": 106010
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.19630752357305e-05,
+      "loss": 0.997,
+      "step": 106020
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.1962313586302496e-05,
+      "loss": 0.9411,
+      "step": 106030
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.19615519368745e-05,
+      "loss": 0.944,
+      "step": 106040
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.196079028744649e-05,
+      "loss": 1.0762,
+      "step": 106050
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.1960028638018495e-05,
+      "loss": 0.9222,
+      "step": 106060
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.195926698859049e-05,
+      "loss": 1.1171,
+      "step": 106070
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.195850533916249e-05,
+      "loss": 0.9134,
+      "step": 106080
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.195774368973449e-05,
+      "loss": 1.0169,
+      "step": 106090
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.195698204030649e-05,
+      "loss": 1.0444,
+      "step": 106100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195622039087849e-05,
+      "loss": 1.0723,
+      "step": 106110
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195545874145049e-05,
+      "loss": 0.9253,
+      "step": 106120
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195469709202249e-05,
+      "loss": 0.9503,
+      "step": 106130
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195393544259448e-05,
+      "loss": 0.8682,
+      "step": 106140
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1953173793166486e-05,
+      "loss": 1.0182,
+      "step": 106150
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1952412143738484e-05,
+      "loss": 1.0162,
+      "step": 106160
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195165049431048e-05,
+      "loss": 1.0622,
+      "step": 106170
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195088884488248e-05,
+      "loss": 0.9928,
+      "step": 106180
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.195012719545448e-05,
+      "loss": 0.9846,
+      "step": 106190
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1949365546026474e-05,
+      "loss": 0.9701,
+      "step": 106200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194860389659848e-05,
+      "loss": 0.9608,
+      "step": 106210
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1947842247170475e-05,
+      "loss": 1.1171,
+      "step": 106220
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194708059774247e-05,
+      "loss": 0.9527,
+      "step": 106230
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194631894831447e-05,
+      "loss": 0.9048,
+      "step": 106240
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1945557298886474e-05,
+      "loss": 0.9542,
+      "step": 106250
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1944795649458465e-05,
+      "loss": 1.024,
+      "step": 106260
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194403400003047e-05,
+      "loss": 0.9402,
+      "step": 106270
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194327235060247e-05,
+      "loss": 0.983,
+      "step": 106280
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1942510701174464e-05,
+      "loss": 1.0787,
+      "step": 106290
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194174905174646e-05,
+      "loss": 0.9244,
+      "step": 106300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.194098740231846e-05,
+      "loss": 0.9386,
+      "step": 106310
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.1940225752890463e-05,
+      "loss": 1.0498,
+      "step": 106320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193946410346246e-05,
+      "loss": 0.9723,
+      "step": 106330
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193870245403446e-05,
+      "loss": 0.9515,
+      "step": 106340
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1937940804606456e-05,
+      "loss": 0.9806,
+      "step": 106350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193717915517846e-05,
+      "loss": 0.9779,
+      "step": 106360
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193641750575045e-05,
+      "loss": 0.9575,
+      "step": 106370
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1935655856322455e-05,
+      "loss": 1.0043,
+      "step": 106380
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193489420689445e-05,
+      "loss": 1.0208,
+      "step": 106390
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193413255746645e-05,
+      "loss": 0.9769,
+      "step": 106400
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193337090803845e-05,
+      "loss": 1.0078,
+      "step": 106410
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193260925861045e-05,
+      "loss": 0.9796,
+      "step": 106420
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193184760918244e-05,
+      "loss": 0.9227,
+      "step": 106430
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.193108595975445e-05,
+      "loss": 0.9537,
+      "step": 106440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1930324310326444e-05,
+      "loss": 0.9434,
+      "step": 106450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192956266089844e-05,
+      "loss": 0.9024,
+      "step": 106460
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192880101147044e-05,
+      "loss": 1.0978,
+      "step": 106470
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192803936204244e-05,
+      "loss": 0.8951,
+      "step": 106480
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192727771261444e-05,
+      "loss": 0.9072,
+      "step": 106490
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192651606318644e-05,
+      "loss": 0.994,
+      "step": 106500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1925754413758436e-05,
+      "loss": 0.8411,
+      "step": 106510
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192499276433043e-05,
+      "loss": 0.9745,
+      "step": 106520
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.192423111490244e-05,
+      "loss": 0.944,
+      "step": 106530
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1923469465474435e-05,
+      "loss": 1.0035,
+      "step": 106540
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.192270781604643e-05,
+      "loss": 1.0882,
+      "step": 106550
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.192194616661843e-05,
+      "loss": 1.0559,
+      "step": 106560
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1921184517190434e-05,
+      "loss": 0.9751,
+      "step": 106570
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1920422867762425e-05,
+      "loss": 0.9357,
+      "step": 106580
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191966121833443e-05,
+      "loss": 1.0518,
+      "step": 106590
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1918899568906426e-05,
+      "loss": 1.0614,
+      "step": 106600
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1918137919478424e-05,
+      "loss": 1.0051,
+      "step": 106610
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191737627005042e-05,
+      "loss": 1.0993,
+      "step": 106620
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1916614620622426e-05,
+      "loss": 0.9738,
+      "step": 106630
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1915852971194416e-05,
+      "loss": 0.9805,
+      "step": 106640
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191509132176642e-05,
+      "loss": 0.8973,
+      "step": 106650
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191432967233842e-05,
+      "loss": 0.9338,
+      "step": 106660
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1913568022910415e-05,
+      "loss": 0.9459,
+      "step": 106670
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191280637348241e-05,
+      "loss": 0.9547,
+      "step": 106680
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191204472405442e-05,
+      "loss": 0.9587,
+      "step": 106690
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1911283074626415e-05,
+      "loss": 1.146,
+      "step": 106700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.191052142519841e-05,
+      "loss": 0.9516,
+      "step": 106710
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1909759775770416e-05,
+      "loss": 0.871,
+      "step": 106720
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.190899812634241e-05,
+      "loss": 1.1172,
+      "step": 106730
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.190823647691441e-05,
+      "loss": 0.9646,
+      "step": 106740
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.190747482748641e-05,
+      "loss": 1.0455,
+      "step": 106750
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.1906713178058406e-05,
+      "loss": 1.0428,
+      "step": 106760
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1905951528630404e-05,
+      "loss": 0.9282,
+      "step": 106770
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.19051898792024e-05,
+      "loss": 1.0323,
+      "step": 106780
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.19044282297744e-05,
+      "loss": 1.0683,
+      "step": 106790
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.19036665803464e-05,
+      "loss": 1.1461,
+      "step": 106800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1902904930918394e-05,
+      "loss": 0.9468,
+      "step": 106810
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.19021432814904e-05,
+      "loss": 0.9254,
+      "step": 106820
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1901381632062395e-05,
+      "loss": 0.9567,
+      "step": 106830
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.190061998263439e-05,
+      "loss": 1.0042,
+      "step": 106840
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189985833320639e-05,
+      "loss": 0.9797,
+      "step": 106850
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1899096683778394e-05,
+      "loss": 0.9711,
+      "step": 106860
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189833503435039e-05,
+      "loss": 0.9753,
+      "step": 106870
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189757338492239e-05,
+      "loss": 1.0597,
+      "step": 106880
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189681173549439e-05,
+      "loss": 1.029,
+      "step": 106890
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1896050086066384e-05,
+      "loss": 0.9026,
+      "step": 106900
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189528843663839e-05,
+      "loss": 0.9522,
+      "step": 106910
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1894526787210386e-05,
+      "loss": 1.0517,
+      "step": 106920
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1893765137782384e-05,
+      "loss": 1.0638,
+      "step": 106930
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189300348835438e-05,
+      "loss": 0.9778,
+      "step": 106940
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1892241838926385e-05,
+      "loss": 1.0062,
+      "step": 106950
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1891480189498376e-05,
+      "loss": 1.1378,
+      "step": 106960
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.189071854007038e-05,
+      "loss": 0.9012,
+      "step": 106970
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.188995689064238e-05,
+      "loss": 0.8695,
+      "step": 106980
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.1889195241214375e-05,
+      "loss": 0.944,
+      "step": 106990
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188843359178637e-05,
+      "loss": 0.9779,
+      "step": 107000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188767194235838e-05,
+      "loss": 1.1201,
+      "step": 107010
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188691029293037e-05,
+      "loss": 1.0166,
+      "step": 107020
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188614864350237e-05,
+      "loss": 1.0863,
+      "step": 107030
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188538699407437e-05,
+      "loss": 0.9969,
+      "step": 107040
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188462534464637e-05,
+      "loss": 1.0112,
+      "step": 107050
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.1883863695218364e-05,
+      "loss": 0.9239,
+      "step": 107060
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188310204579037e-05,
+      "loss": 0.9578,
+      "step": 107070
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.1882340396362366e-05,
+      "loss": 1.0069,
+      "step": 107080
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188157874693436e-05,
+      "loss": 1.07,
+      "step": 107090
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188081709750636e-05,
+      "loss": 0.998,
+      "step": 107100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.188005544807836e-05,
+      "loss": 0.9978,
+      "step": 107110
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.187929379865036e-05,
+      "loss": 0.9241,
+      "step": 107120
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.187853214922236e-05,
+      "loss": 1.0297,
+      "step": 107130
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.187777049979436e-05,
+      "loss": 0.9298,
+      "step": 107140
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.1877008850366355e-05,
+      "loss": 1.022,
+      "step": 107150
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.187624720093836e-05,
+      "loss": 0.928,
+      "step": 107160
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.187548555151035e-05,
+      "loss": 0.9287,
+      "step": 107170
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.1874723902082354e-05,
+      "loss": 1.0845,
+      "step": 107180
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.187396225265435e-05,
+      "loss": 0.9367,
+      "step": 107190
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.187320060322635e-05,
+      "loss": 1.121,
+      "step": 107200
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1872438953798346e-05,
+      "loss": 0.9869,
+      "step": 107210
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.187167730437035e-05,
+      "loss": 0.9736,
+      "step": 107220
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.187091565494234e-05,
+      "loss": 1.0827,
+      "step": 107230
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1870154005514346e-05,
+      "loss": 1.0042,
+      "step": 107240
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186939235608634e-05,
+      "loss": 1.1323,
+      "step": 107250
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186863070665834e-05,
+      "loss": 1.2197,
+      "step": 107260
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186786905723034e-05,
+      "loss": 0.9063,
+      "step": 107270
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1867107407802335e-05,
+      "loss": 0.9612,
+      "step": 107280
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186634575837434e-05,
+      "loss": 1.0504,
+      "step": 107290
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186558410894634e-05,
+      "loss": 1.0283,
+      "step": 107300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1864822459518335e-05,
+      "loss": 1.0138,
+      "step": 107310
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186406081009033e-05,
+      "loss": 1.0588,
+      "step": 107320
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1863299160662336e-05,
+      "loss": 0.9299,
+      "step": 107330
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186253751123433e-05,
+      "loss": 0.965,
+      "step": 107340
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186177586180633e-05,
+      "loss": 1.119,
+      "step": 107350
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.186101421237833e-05,
+      "loss": 0.9976,
+      "step": 107360
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1860252562950326e-05,
+      "loss": 1.0685,
+      "step": 107370
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1859490913522324e-05,
+      "loss": 1.0036,
+      "step": 107380
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.185872926409433e-05,
+      "loss": 1.0319,
+      "step": 107390
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.185796761466632e-05,
+      "loss": 1.005,
+      "step": 107400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.185720596523832e-05,
+      "loss": 0.9737,
+      "step": 107410
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.185644431581032e-05,
+      "loss": 1.0337,
+      "step": 107420
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.185568266638232e-05,
+      "loss": 1.0951,
+      "step": 107430
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.1854921016954315e-05,
+      "loss": 0.9335,
+      "step": 107440
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.185415936752632e-05,
+      "loss": 1.003,
+      "step": 107450
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.185339771809832e-05,
+      "loss": 0.9268,
+      "step": 107460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.1852636068670314e-05,
+      "loss": 0.8721,
+      "step": 107470
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.185187441924231e-05,
+      "loss": 0.9485,
+      "step": 107480
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.185111276981431e-05,
+      "loss": 1.2242,
+      "step": 107490
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.1850351120386314e-05,
+      "loss": 0.9306,
+      "step": 107500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.184958947095831e-05,
+      "loss": 1.0727,
+      "step": 107510
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.184882782153031e-05,
+      "loss": 1.0141,
+      "step": 107520
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.1848066172102306e-05,
+      "loss": 0.8763,
+      "step": 107530
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.184730452267431e-05,
+      "loss": 0.9813,
+      "step": 107540
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.18465428732463e-05,
+      "loss": 0.9521,
+      "step": 107550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.1845781223818305e-05,
+      "loss": 1.0154,
+      "step": 107560
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.18450195743903e-05,
+      "loss": 1.1039,
+      "step": 107570
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.18442579249623e-05,
+      "loss": 1.022,
+      "step": 107580
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.18434962755343e-05,
+      "loss": 0.9452,
+      "step": 107590
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.18427346261063e-05,
+      "loss": 0.9699,
+      "step": 107600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.184197297667829e-05,
+      "loss": 1.12,
+      "step": 107610
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.18412113272503e-05,
+      "loss": 0.9074,
+      "step": 107620
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.1840449677822294e-05,
+      "loss": 0.9746,
+      "step": 107630
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183968802839429e-05,
+      "loss": 0.9841,
+      "step": 107640
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183892637896629e-05,
+      "loss": 0.9761,
+      "step": 107650
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1838164729538293e-05,
+      "loss": 1.0963,
+      "step": 107660
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183740308011029e-05,
+      "loss": 1.0073,
+      "step": 107670
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183664143068229e-05,
+      "loss": 0.9853,
+      "step": 107680
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1835879781254286e-05,
+      "loss": 0.9238,
+      "step": 107690
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183511813182628e-05,
+      "loss": 0.9124,
+      "step": 107700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183435648239829e-05,
+      "loss": 0.996,
+      "step": 107710
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183359483297028e-05,
+      "loss": 0.9682,
+      "step": 107720
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183283318354228e-05,
+      "loss": 1.0155,
+      "step": 107730
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183207153411428e-05,
+      "loss": 0.9581,
+      "step": 107740
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.183130988468628e-05,
+      "loss": 1.0281,
+      "step": 107750
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1830548235258275e-05,
+      "loss": 0.9598,
+      "step": 107760
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.182978658583028e-05,
+      "loss": 1.1453,
+      "step": 107770
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.182902493640227e-05,
+      "loss": 1.0488,
+      "step": 107780
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1828263286974274e-05,
+      "loss": 1.0353,
+      "step": 107790
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.182750163754627e-05,
+      "loss": 1.0014,
+      "step": 107800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.182673998811827e-05,
+      "loss": 0.9966,
+      "step": 107810
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1825978338690266e-05,
+      "loss": 0.8954,
+      "step": 107820
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.182521668926227e-05,
+      "loss": 0.9276,
+      "step": 107830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.182445503983427e-05,
+      "loss": 1.016,
+      "step": 107840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1823693390406266e-05,
+      "loss": 0.8985,
+      "step": 107850
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.182293174097826e-05,
+      "loss": 0.922,
+      "step": 107860
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.182217009155026e-05,
+      "loss": 0.9006,
+      "step": 107870
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.1821408442122265e-05,
+      "loss": 0.9905,
+      "step": 107880
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.182064679269426e-05,
+      "loss": 0.9807,
+      "step": 107890
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181988514326626e-05,
+      "loss": 0.9576,
+      "step": 107900
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181912349383826e-05,
+      "loss": 1.0309,
+      "step": 107910
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181836184441026e-05,
+      "loss": 0.9941,
+      "step": 107920
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181760019498225e-05,
+      "loss": 0.9715,
+      "step": 107930
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.1816838545554256e-05,
+      "loss": 0.9523,
+      "step": 107940
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.1816076896126254e-05,
+      "loss": 0.9683,
+      "step": 107950
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181531524669825e-05,
+      "loss": 0.9879,
+      "step": 107960
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181455359727025e-05,
+      "loss": 1.1389,
+      "step": 107970
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181379194784225e-05,
+      "loss": 0.9818,
+      "step": 107980
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.1813030298414244e-05,
+      "loss": 0.9723,
+      "step": 107990
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181226864898625e-05,
+      "loss": 0.987,
+      "step": 108000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.1811506999558245e-05,
+      "loss": 1.0415,
+      "step": 108010
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.181074535013024e-05,
+      "loss": 1.0845,
+      "step": 108020
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.180998370070224e-05,
+      "loss": 0.9491,
+      "step": 108030
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.1809222051274245e-05,
+      "loss": 1.0089,
+      "step": 108040
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.180846040184624e-05,
+      "loss": 1.065,
+      "step": 108050
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.180769875241824e-05,
+      "loss": 0.9471,
+      "step": 108060
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.180693710299024e-05,
+      "loss": 1.0247,
+      "step": 108070
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1806175453562234e-05,
+      "loss": 0.9776,
+      "step": 108080
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.180541380413424e-05,
+      "loss": 0.9411,
+      "step": 108090
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1804652154706236e-05,
+      "loss": 0.9122,
+      "step": 108100
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1803890505278234e-05,
+      "loss": 0.9507,
+      "step": 108110
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.180312885585023e-05,
+      "loss": 0.9115,
+      "step": 108120
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1802367206422235e-05,
+      "loss": 0.9307,
+      "step": 108130
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1801605556994226e-05,
+      "loss": 1.1172,
+      "step": 108140
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.180084390756623e-05,
+      "loss": 1.0611,
+      "step": 108150
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.180008225813822e-05,
+      "loss": 0.9897,
+      "step": 108160
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1799320608710225e-05,
+      "loss": 1.019,
+      "step": 108170
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179855895928222e-05,
+      "loss": 0.8796,
+      "step": 108180
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179779730985422e-05,
+      "loss": 1.0223,
+      "step": 108190
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179703566042622e-05,
+      "loss": 0.9914,
+      "step": 108200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179627401099822e-05,
+      "loss": 0.9176,
+      "step": 108210
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179551236157022e-05,
+      "loss": 0.9446,
+      "step": 108220
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179475071214222e-05,
+      "loss": 1.0058,
+      "step": 108230
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1793989062714214e-05,
+      "loss": 1.0502,
+      "step": 108240
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179322741328621e-05,
+      "loss": 1.1179,
+      "step": 108250
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1792465763858216e-05,
+      "loss": 0.8864,
+      "step": 108260
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1791704114430213e-05,
+      "loss": 0.9474,
+      "step": 108270
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179094246500221e-05,
+      "loss": 0.9783,
+      "step": 108280
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.179018081557421e-05,
+      "loss": 1.013,
+      "step": 108290
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.178941916614621e-05,
+      "loss": 0.9823,
+      "step": 108300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.17886575167182e-05,
+      "loss": 0.9833,
+      "step": 108310
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.178789586729021e-05,
+      "loss": 0.9817,
+      "step": 108320
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1787134217862205e-05,
+      "loss": 0.9458,
+      "step": 108330
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.17863725684342e-05,
+      "loss": 0.9583,
+      "step": 108340
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.17856109190062e-05,
+      "loss": 1.0206,
+      "step": 108350
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1784849269578204e-05,
+      "loss": 0.9397,
+      "step": 108360
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1784087620150195e-05,
+      "loss": 1.0863,
+      "step": 108370
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.17833259707222e-05,
+      "loss": 0.9251,
+      "step": 108380
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1782564321294197e-05,
+      "loss": 1.1561,
+      "step": 108390
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1781802671866194e-05,
+      "loss": 0.9846,
+      "step": 108400
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.178104102243819e-05,
+      "loss": 1.088,
+      "step": 108410
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1780279373010196e-05,
+      "loss": 0.9489,
+      "step": 108420
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.177951772358219e-05,
+      "loss": 0.8895,
+      "step": 108430
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.177875607415419e-05,
+      "loss": 0.9158,
+      "step": 108440
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.177799442472619e-05,
+      "loss": 1.1248,
+      "step": 108450
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1777232775298186e-05,
+      "loss": 0.9589,
+      "step": 108460
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.177647112587019e-05,
+      "loss": 1.0551,
+      "step": 108470
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.177570947644219e-05,
+      "loss": 0.9719,
+      "step": 108480
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1774947827014185e-05,
+      "loss": 0.9277,
+      "step": 108490
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.177418617758618e-05,
+      "loss": 0.932,
+      "step": 108500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1773424528158187e-05,
+      "loss": 0.9414,
+      "step": 108510
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.177266287873018e-05,
+      "loss": 0.9507,
+      "step": 108520
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.177190122930218e-05,
+      "loss": 1.016,
+      "step": 108530
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.177113957987418e-05,
+      "loss": 1.0593,
+      "step": 108540
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1770377930446176e-05,
+      "loss": 0.9783,
+      "step": 108550
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1769616281018174e-05,
+      "loss": 0.9328,
+      "step": 108560
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176885463159018e-05,
+      "loss": 1.0151,
+      "step": 108570
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176809298216217e-05,
+      "loss": 0.8806,
+      "step": 108580
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176733133273417e-05,
+      "loss": 0.9246,
+      "step": 108590
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176656968330617e-05,
+      "loss": 1.0237,
+      "step": 108600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176580803387817e-05,
+      "loss": 1.0594,
+      "step": 108610
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1765046384450165e-05,
+      "loss": 1.0137,
+      "step": 108620
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176428473502216e-05,
+      "loss": 0.9339,
+      "step": 108630
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176352308559417e-05,
+      "loss": 1.0224,
+      "step": 108640
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1762761436166165e-05,
+      "loss": 1.0583,
+      "step": 108650
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176199978673816e-05,
+      "loss": 0.9321,
+      "step": 108660
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.176123813731016e-05,
+      "loss": 0.9584,
+      "step": 108670
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1760476487882164e-05,
+      "loss": 0.9984,
+      "step": 108680
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1759714838454154e-05,
+      "loss": 1.0401,
+      "step": 108690
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.175895318902616e-05,
+      "loss": 0.9362,
+      "step": 108700
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1758191539598156e-05,
+      "loss": 1.2775,
+      "step": 108710
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1757429890170154e-05,
+      "loss": 1.1312,
+      "step": 108720
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.175666824074215e-05,
+      "loss": 1.043,
+      "step": 108730
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1755906591314155e-05,
+      "loss": 0.9998,
+      "step": 108740
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1755144941886146e-05,
+      "loss": 0.9922,
+      "step": 108750
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.175438329245815e-05,
+      "loss": 0.9362,
+      "step": 108760
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.175362164303015e-05,
+      "loss": 0.9027,
+      "step": 108770
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1752859993602145e-05,
+      "loss": 0.9762,
+      "step": 108780
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.175209834417414e-05,
+      "loss": 1.0082,
+      "step": 108790
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.175133669474615e-05,
+      "loss": 0.9792,
+      "step": 108800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.175057504531814e-05,
+      "loss": 1.0066,
+      "step": 108810
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174981339589014e-05,
+      "loss": 0.9523,
+      "step": 108820
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174905174646214e-05,
+      "loss": 1.0719,
+      "step": 108830
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174829009703414e-05,
+      "loss": 1.1395,
+      "step": 108840
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174752844760614e-05,
+      "loss": 0.982,
+      "step": 108850
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174676679817814e-05,
+      "loss": 0.9151,
+      "step": 108860
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1746005148750136e-05,
+      "loss": 0.9879,
+      "step": 108870
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1745243499322133e-05,
+      "loss": 0.9622,
+      "step": 108880
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174448184989414e-05,
+      "loss": 0.9121,
+      "step": 108890
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174372020046613e-05,
+      "loss": 0.9741,
+      "step": 108900
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174295855103813e-05,
+      "loss": 0.9163,
+      "step": 108910
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174219690161013e-05,
+      "loss": 0.9894,
+      "step": 108920
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174143525218213e-05,
+      "loss": 1.0426,
+      "step": 108930
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1740673602754125e-05,
+      "loss": 0.9358,
+      "step": 108940
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.173991195332613e-05,
+      "loss": 1.0291,
+      "step": 108950
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173915030389812e-05,
+      "loss": 1.007,
+      "step": 108960
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1738388654470124e-05,
+      "loss": 1.0256,
+      "step": 108970
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173762700504212e-05,
+      "loss": 1.0497,
+      "step": 108980
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173686535561412e-05,
+      "loss": 0.9999,
+      "step": 108990
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173610370618612e-05,
+      "loss": 0.9251,
+      "step": 109000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173534205675812e-05,
+      "loss": 0.9546,
+      "step": 109010
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173458040733012e-05,
+      "loss": 0.9947,
+      "step": 109020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1733818757902116e-05,
+      "loss": 1.1501,
+      "step": 109030
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173305710847411e-05,
+      "loss": 1.064,
+      "step": 109040
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173229545904611e-05,
+      "loss": 0.9659,
+      "step": 109050
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1731533809618115e-05,
+      "loss": 0.9711,
+      "step": 109060
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1730772160190106e-05,
+      "loss": 1.1551,
+      "step": 109070
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.173001051076211e-05,
+      "loss": 0.9968,
+      "step": 109080
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.172924886133411e-05,
+      "loss": 0.9557,
+      "step": 109090
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1728487211906105e-05,
+      "loss": 0.9485,
+      "step": 109100
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.17277255624781e-05,
+      "loss": 0.9663,
+      "step": 109110
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1726963913050107e-05,
+      "loss": 0.9657,
+      "step": 109120
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.17262022636221e-05,
+      "loss": 1.0673,
+      "step": 109130
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.17254406141941e-05,
+      "loss": 0.9201,
+      "step": 109140
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.17246789647661e-05,
+      "loss": 1.0496,
+      "step": 109150
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1723917315338096e-05,
+      "loss": 0.9223,
+      "step": 109160
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.1723155665910094e-05,
+      "loss": 0.9926,
+      "step": 109170
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.17223940164821e-05,
+      "loss": 0.9854,
+      "step": 109180
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.172163236705409e-05,
+      "loss": 0.9949,
+      "step": 109190
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.172087071762609e-05,
+      "loss": 0.9275,
+      "step": 109200
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.172010906819809e-05,
+      "loss": 0.9459,
+      "step": 109210
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171934741877009e-05,
+      "loss": 1.0358,
+      "step": 109220
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171858576934209e-05,
+      "loss": 0.9287,
+      "step": 109230
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171782411991409e-05,
+      "loss": 1.1003,
+      "step": 109240
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171706247048609e-05,
+      "loss": 1.0734,
+      "step": 109250
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.1716300821058085e-05,
+      "loss": 1.0831,
+      "step": 109260
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171553917163009e-05,
+      "loss": 0.983,
+      "step": 109270
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171477752220208e-05,
+      "loss": 1.039,
+      "step": 109280
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.1714015872774084e-05,
+      "loss": 1.055,
+      "step": 109290
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171325422334608e-05,
+      "loss": 0.995,
+      "step": 109300
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171249257391808e-05,
+      "loss": 0.9784,
+      "step": 109310
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.1711730924490076e-05,
+      "loss": 0.9943,
+      "step": 109320
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171096927506208e-05,
+      "loss": 1.1144,
+      "step": 109330
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.171020762563407e-05,
+      "loss": 1.073,
+      "step": 109340
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.1709445976206075e-05,
+      "loss": 0.9785,
+      "step": 109350
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.170868432677807e-05,
+      "loss": 0.9483,
+      "step": 109360
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.170792267735007e-05,
+      "loss": 1.0259,
+      "step": 109370
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.170716102792207e-05,
+      "loss": 1.0095,
+      "step": 109380
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.170639937849407e-05,
+      "loss": 1.0823,
+      "step": 109390
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.170563772906606e-05,
+      "loss": 0.9953,
+      "step": 109400
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.170487607963807e-05,
+      "loss": 0.8954,
+      "step": 109410
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1704114430210064e-05,
+      "loss": 1.0235,
+      "step": 109420
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.170335278078206e-05,
+      "loss": 0.9814,
+      "step": 109430
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1702591131354066e-05,
+      "loss": 1.123,
+      "step": 109440
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1701829481926064e-05,
+      "loss": 1.0117,
+      "step": 109450
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.170106783249806e-05,
+      "loss": 1.0176,
+      "step": 109460
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.170030618307006e-05,
+      "loss": 1.0307,
+      "step": 109470
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169954453364206e-05,
+      "loss": 0.97,
+      "step": 109480
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1698782884214053e-05,
+      "loss": 0.9523,
+      "step": 109490
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 0.6221529525860532,
+      "eval_em": 0.007379375591296121,
+      "eval_f1": 0.007379375591296121,
+      "eval_loss": 0.9185465574264526,
+      "eval_runtime": 1010.8517,
+      "eval_samples_per_second": 10.457,
+      "eval_steps_per_second": 1.308,
+      "eval_wer": 0.9926206244087039,
+      "step": 109498
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169802123478606e-05,
+      "loss": 0.9439,
+      "step": 109500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1697259585358055e-05,
+      "loss": 1.0604,
+      "step": 109510
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169649793593005e-05,
+      "loss": 0.9846,
+      "step": 109520
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169573628650205e-05,
+      "loss": 0.947,
+      "step": 109530
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169497463707405e-05,
+      "loss": 0.9131,
+      "step": 109540
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1694212987646045e-05,
+      "loss": 0.8954,
+      "step": 109550
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169345133821805e-05,
+      "loss": 0.9481,
+      "step": 109560
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169268968879004e-05,
+      "loss": 1.0711,
+      "step": 109570
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.1691928039362044e-05,
+      "loss": 0.9986,
+      "step": 109580
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169116638993404e-05,
+      "loss": 0.9466,
+      "step": 109590
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.169040474050604e-05,
+      "loss": 0.9963,
+      "step": 109600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168964309107804e-05,
+      "loss": 1.0027,
+      "step": 109610
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168888144165004e-05,
+      "loss": 0.9988,
+      "step": 109620
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168811979222204e-05,
+      "loss": 1.022,
+      "step": 109630
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.1687358142794036e-05,
+      "loss": 0.8803,
+      "step": 109640
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168659649336604e-05,
+      "loss": 0.9009,
+      "step": 109650
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168583484393803e-05,
+      "loss": 1.031,
+      "step": 109660
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.1685073194510035e-05,
+      "loss": 0.9758,
+      "step": 109670
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168431154508203e-05,
+      "loss": 0.9099,
+      "step": 109680
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168354989565403e-05,
+      "loss": 1.0329,
+      "step": 109690
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168278824622603e-05,
+      "loss": 0.8441,
+      "step": 109700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168202659679803e-05,
+      "loss": 1.0267,
+      "step": 109710
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.168126494737002e-05,
+      "loss": 0.9883,
+      "step": 109720
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.1680503297942027e-05,
+      "loss": 0.8761,
+      "step": 109730
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.1679741648514024e-05,
+      "loss": 0.9327,
+      "step": 109740
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.167897999908602e-05,
+      "loss": 0.9616,
+      "step": 109750
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.167821834965802e-05,
+      "loss": 0.9652,
+      "step": 109760
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.167745670023002e-05,
+      "loss": 0.9045,
+      "step": 109770
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.1676695050802014e-05,
+      "loss": 0.9052,
+      "step": 109780
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.167593340137402e-05,
+      "loss": 1.0296,
+      "step": 109790
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.1675171751946016e-05,
+      "loss": 0.9594,
+      "step": 109800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.167441010251801e-05,
+      "loss": 0.9572,
+      "step": 109810
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.167364845309002e-05,
+      "loss": 0.913,
+      "step": 109820
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1672886803662015e-05,
+      "loss": 1.0159,
+      "step": 109830
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.167212515423401e-05,
+      "loss": 1.0515,
+      "step": 109840
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.167136350480601e-05,
+      "loss": 1.0437,
+      "step": 109850
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1670601855378014e-05,
+      "loss": 0.9243,
+      "step": 109860
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1669840205950005e-05,
+      "loss": 1.0268,
+      "step": 109870
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.166907855652201e-05,
+      "loss": 0.9241,
+      "step": 109880
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1668316907094006e-05,
+      "loss": 0.9639,
+      "step": 109890
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1667555257666004e-05,
+      "loss": 1.0387,
+      "step": 109900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1666793608238e-05,
+      "loss": 0.9685,
+      "step": 109910
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1666031958810006e-05,
+      "loss": 0.9385,
+      "step": 109920
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1665270309381996e-05,
+      "loss": 1.0417,
+      "step": 109930
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1664508659954e-05,
+      "loss": 0.8954,
+      "step": 109940
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1663747010526e-05,
+      "loss": 0.9645,
+      "step": 109950
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1662985361097995e-05,
+      "loss": 0.9177,
+      "step": 109960
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.166222371166999e-05,
+      "loss": 0.8148,
+      "step": 109970
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.1661462062242e-05,
+      "loss": 0.9694,
+      "step": 109980
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.166070041281399e-05,
+      "loss": 0.9437,
+      "step": 109990
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.165993876338599e-05,
+      "loss": 0.8939,
+      "step": 110000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.165917711395799e-05,
+      "loss": 1.0471,
+      "step": 110010
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.165841546452999e-05,
+      "loss": 0.9159,
+      "step": 110020
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.165765381510199e-05,
+      "loss": 0.9041,
+      "step": 110030
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.165689216567398e-05,
+      "loss": 1.0549,
+      "step": 110040
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1656130516245986e-05,
+      "loss": 0.8452,
+      "step": 110050
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1655368866817984e-05,
+      "loss": 0.9761,
+      "step": 110060
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.165460721738998e-05,
+      "loss": 0.9372,
+      "step": 110070
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.165384556796198e-05,
+      "loss": 0.9987,
+      "step": 110080
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.165308391853398e-05,
+      "loss": 0.9014,
+      "step": 110090
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1652322269105973e-05,
+      "loss": 0.8295,
+      "step": 110100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.165156061967798e-05,
+      "loss": 1.0277,
+      "step": 110110
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1650798970249975e-05,
+      "loss": 0.9459,
+      "step": 110120
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.165003732082197e-05,
+      "loss": 0.9406,
+      "step": 110130
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.164927567139397e-05,
+      "loss": 0.9032,
+      "step": 110140
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1648514021965974e-05,
+      "loss": 0.7935,
+      "step": 110150
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1647752372537965e-05,
+      "loss": 0.8073,
+      "step": 110160
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.164699072310997e-05,
+      "loss": 0.924,
+      "step": 110170
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.164622907368197e-05,
+      "loss": 0.9023,
+      "step": 110180
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1645467424253964e-05,
+      "loss": 1.0524,
+      "step": 110190
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.164470577482596e-05,
+      "loss": 0.8802,
+      "step": 110200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1643944125397966e-05,
+      "loss": 0.9926,
+      "step": 110210
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1643182475969963e-05,
+      "loss": 0.9787,
+      "step": 110220
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.164242082654196e-05,
+      "loss": 0.8589,
+      "step": 110230
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1641659177113965e-05,
+      "loss": 0.9117,
+      "step": 110240
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1640897527685956e-05,
+      "loss": 0.9144,
+      "step": 110250
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.164013587825796e-05,
+      "loss": 0.8552,
+      "step": 110260
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163937422882996e-05,
+      "loss": 0.8827,
+      "step": 110270
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1638612579401955e-05,
+      "loss": 0.9221,
+      "step": 110280
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163785092997395e-05,
+      "loss": 0.9588,
+      "step": 110290
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163708928054596e-05,
+      "loss": 0.9669,
+      "step": 110300
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163632763111795e-05,
+      "loss": 1.1402,
+      "step": 110310
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163556598168995e-05,
+      "loss": 0.9113,
+      "step": 110320
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163480433226195e-05,
+      "loss": 0.9535,
+      "step": 110330
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1634042682833947e-05,
+      "loss": 0.9813,
+      "step": 110340
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1633281033405944e-05,
+      "loss": 1.032,
+      "step": 110350
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163251938397795e-05,
+      "loss": 0.98,
+      "step": 110360
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163175773454994e-05,
+      "loss": 0.9224,
+      "step": 110370
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163099608512194e-05,
+      "loss": 0.9412,
+      "step": 110380
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.163023443569394e-05,
+      "loss": 0.852,
+      "step": 110390
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.162947278626594e-05,
+      "loss": 1.1584,
+      "step": 110400
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1628711136837936e-05,
+      "loss": 0.9977,
+      "step": 110410
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.162794948740994e-05,
+      "loss": 1.1382,
+      "step": 110420
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.162718783798194e-05,
+      "loss": 1.0877,
+      "step": 110430
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1626426188553935e-05,
+      "loss": 1.0987,
+      "step": 110440
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.162566453912593e-05,
+      "loss": 0.9764,
+      "step": 110450
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.162490288969793e-05,
+      "loss": 0.9878,
+      "step": 110460
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1624141240269934e-05,
+      "loss": 0.9591,
+      "step": 110470
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.1623379590841925e-05,
+      "loss": 0.9889,
+      "step": 110480
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.162261794141393e-05,
+      "loss": 1.0555,
+      "step": 110490
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1621856291985926e-05,
+      "loss": 0.9856,
+      "step": 110500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1621094642557924e-05,
+      "loss": 0.931,
+      "step": 110510
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.162033299312992e-05,
+      "loss": 0.8467,
+      "step": 110520
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1619571343701926e-05,
+      "loss": 0.9936,
+      "step": 110530
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1618809694273916e-05,
+      "loss": 0.9914,
+      "step": 110540
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161804804484592e-05,
+      "loss": 0.9,
+      "step": 110550
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161728639541792e-05,
+      "loss": 0.8575,
+      "step": 110560
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1616524745989915e-05,
+      "loss": 0.9455,
+      "step": 110570
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161576309656191e-05,
+      "loss": 0.9368,
+      "step": 110580
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161500144713392e-05,
+      "loss": 0.9979,
+      "step": 110590
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1614239797705915e-05,
+      "loss": 0.8517,
+      "step": 110600
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161347814827791e-05,
+      "loss": 0.9115,
+      "step": 110610
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1612716498849916e-05,
+      "loss": 0.9707,
+      "step": 110620
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161195484942191e-05,
+      "loss": 0.906,
+      "step": 110630
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161119319999391e-05,
+      "loss": 0.9036,
+      "step": 110640
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.161043155056591e-05,
+      "loss": 0.9885,
+      "step": 110650
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1609669901137906e-05,
+      "loss": 0.8841,
+      "step": 110660
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.1608908251709904e-05,
+      "loss": 1.0625,
+      "step": 110670
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.160814660228191e-05,
+      "loss": 1.0414,
+      "step": 110680
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.16073849528539e-05,
+      "loss": 0.9672,
+      "step": 110690
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.16066233034259e-05,
+      "loss": 0.8808,
+      "step": 110700
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.16058616539979e-05,
+      "loss": 0.9225,
+      "step": 110710
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.16051000045699e-05,
+      "loss": 0.8162,
+      "step": 110720
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.1604338355141895e-05,
+      "loss": 1.0878,
+      "step": 110730
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.16035767057139e-05,
+      "loss": 0.8652,
+      "step": 110740
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.160281505628589e-05,
+      "loss": 1.023,
+      "step": 110750
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.1602053406857894e-05,
+      "loss": 0.8981,
+      "step": 110760
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.160129175742989e-05,
+      "loss": 0.9017,
+      "step": 110770
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.160053010800189e-05,
+      "loss": 0.9245,
+      "step": 110780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159976845857389e-05,
+      "loss": 0.9145,
+      "step": 110790
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159900680914589e-05,
+      "loss": 1.0537,
+      "step": 110800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159824515971789e-05,
+      "loss": 0.9875,
+      "step": 110810
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.1597483510289886e-05,
+      "loss": 0.9998,
+      "step": 110820
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159672186086189e-05,
+      "loss": 1.0922,
+      "step": 110830
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159596021143388e-05,
+      "loss": 1.0015,
+      "step": 110840
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.1595198562005885e-05,
+      "loss": 1.094,
+      "step": 110850
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159443691257788e-05,
+      "loss": 0.9196,
+      "step": 110860
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159367526314988e-05,
+      "loss": 0.9243,
+      "step": 110870
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159291361372188e-05,
+      "loss": 0.971,
+      "step": 110880
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159215196429388e-05,
+      "loss": 0.8249,
+      "step": 110890
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159139031486587e-05,
+      "loss": 0.8186,
+      "step": 110900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.159062866543788e-05,
+      "loss": 1.0194,
+      "step": 110910
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.158986701600987e-05,
+      "loss": 1.0105,
+      "step": 110920
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158910536658187e-05,
+      "loss": 0.9187,
+      "step": 110930
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158834371715387e-05,
+      "loss": 0.8617,
+      "step": 110940
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1587582067725867e-05,
+      "loss": 0.9938,
+      "step": 110950
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1586820418297864e-05,
+      "loss": 1.1022,
+      "step": 110960
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158605876886987e-05,
+      "loss": 0.916,
+      "step": 110970
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1585297119441866e-05,
+      "loss": 1.045,
+      "step": 110980
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158453547001386e-05,
+      "loss": 0.9358,
+      "step": 110990
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158377382058586e-05,
+      "loss": 0.9618,
+      "step": 111000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158301217115786e-05,
+      "loss": 0.9476,
+      "step": 111010
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158225052172986e-05,
+      "loss": 0.948,
+      "step": 111020
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158148887230186e-05,
+      "loss": 0.9609,
+      "step": 111030
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.158072722287386e-05,
+      "loss": 1.0535,
+      "step": 111040
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1579965573445855e-05,
+      "loss": 0.9586,
+      "step": 111050
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.157920392401786e-05,
+      "loss": 0.9209,
+      "step": 111060
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.157844227458985e-05,
+      "loss": 0.9965,
+      "step": 111070
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1577680625161854e-05,
+      "loss": 0.881,
+      "step": 111080
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.157691897573385e-05,
+      "loss": 0.9301,
+      "step": 111090
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.157615732630585e-05,
+      "loss": 0.9274,
+      "step": 111100
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1575395676877846e-05,
+      "loss": 1.0279,
+      "step": 111110
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.157463402744985e-05,
+      "loss": 0.8589,
+      "step": 111120
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.157387237802184e-05,
+      "loss": 0.9401,
+      "step": 111130
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.1573110728593846e-05,
+      "loss": 1.0124,
+      "step": 111140
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.157234907916584e-05,
+      "loss": 0.9594,
+      "step": 111150
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.157158742973784e-05,
+      "loss": 0.9531,
+      "step": 111160
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.157082578030984e-05,
+      "loss": 0.968,
+      "step": 111170
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.157006413088184e-05,
+      "loss": 1.0014,
+      "step": 111180
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156930248145384e-05,
+      "loss": 0.9361,
+      "step": 111190
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156854083202584e-05,
+      "loss": 0.9888,
+      "step": 111200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156777918259784e-05,
+      "loss": 1.1226,
+      "step": 111210
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156701753316983e-05,
+      "loss": 0.9492,
+      "step": 111220
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.1566255883741836e-05,
+      "loss": 1.0154,
+      "step": 111230
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.1565494234313834e-05,
+      "loss": 0.972,
+      "step": 111240
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156473258488583e-05,
+      "loss": 0.7718,
+      "step": 111250
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156397093545783e-05,
+      "loss": 0.9453,
+      "step": 111260
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156320928602983e-05,
+      "loss": 1.0697,
+      "step": 111270
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.1562447636601824e-05,
+      "loss": 0.9733,
+      "step": 111280
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156168598717383e-05,
+      "loss": 1.0183,
+      "step": 111290
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.1560924337745825e-05,
+      "loss": 0.9029,
+      "step": 111300
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.156016268831782e-05,
+      "loss": 0.866,
+      "step": 111310
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.155940103888982e-05,
+      "loss": 1.0758,
+      "step": 111320
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.1558639389461825e-05,
+      "loss": 0.8579,
+      "step": 111330
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.1557877740033815e-05,
+      "loss": 1.0383,
+      "step": 111340
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.155711609060582e-05,
+      "loss": 1.0165,
+      "step": 111350
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.155635444117782e-05,
+      "loss": 1.0403,
+      "step": 111360
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.1555592791749814e-05,
+      "loss": 0.8979,
+      "step": 111370
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.155483114232181e-05,
+      "loss": 0.977,
+      "step": 111380
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.155406949289381e-05,
+      "loss": 1.0036,
+      "step": 111390
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.1553307843465814e-05,
+      "loss": 1.0135,
+      "step": 111400
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.155254619403781e-05,
+      "loss": 0.9582,
+      "step": 111410
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.155178454460981e-05,
+      "loss": 0.9253,
+      "step": 111420
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.1551022895181806e-05,
+      "loss": 1.019,
+      "step": 111430
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.155026124575381e-05,
+      "loss": 1.1088,
+      "step": 111440
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.15494995963258e-05,
+      "loss": 0.9839,
+      "step": 111450
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.1548737946897805e-05,
+      "loss": 0.8985,
+      "step": 111460
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.15479762974698e-05,
+      "loss": 0.9,
+      "step": 111470
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.15472146480418e-05,
+      "loss": 1.0076,
+      "step": 111480
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.15464529986138e-05,
+      "loss": 1.0352,
+      "step": 111490
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.15456913491858e-05,
+      "loss": 1.0399,
+      "step": 111500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.154492969975779e-05,
+      "loss": 0.8504,
+      "step": 111510
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.15441680503298e-05,
+      "loss": 1.1276,
+      "step": 111520
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.1543406400901794e-05,
+      "loss": 0.8772,
+      "step": 111530
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.154264475147379e-05,
+      "loss": 0.9842,
+      "step": 111540
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.154188310204579e-05,
+      "loss": 0.9615,
+      "step": 111550
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.154112145261779e-05,
+      "loss": 0.9227,
+      "step": 111560
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.154035980318979e-05,
+      "loss": 1.12,
+      "step": 111570
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153959815376179e-05,
+      "loss": 1.0088,
+      "step": 111580
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1538836504333786e-05,
+      "loss": 0.9847,
+      "step": 111590
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153807485490578e-05,
+      "loss": 1.0426,
+      "step": 111600
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153731320547779e-05,
+      "loss": 1.074,
+      "step": 111610
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1536551556049785e-05,
+      "loss": 1.0094,
+      "step": 111620
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153578990662178e-05,
+      "loss": 0.9493,
+      "step": 111630
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153502825719378e-05,
+      "loss": 0.872,
+      "step": 111640
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1534266607765784e-05,
+      "loss": 1.08,
+      "step": 111650
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1533504958337775e-05,
+      "loss": 1.0086,
+      "step": 111660
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153274330890978e-05,
+      "loss": 0.929,
+      "step": 111670
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1531981659481777e-05,
+      "loss": 0.9919,
+      "step": 111680
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1531220010053774e-05,
+      "loss": 0.9819,
+      "step": 111690
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.153045836062577e-05,
+      "loss": 1.0735,
+      "step": 111700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1529696711197776e-05,
+      "loss": 1.005,
+      "step": 111710
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1528935061769766e-05,
+      "loss": 0.8877,
+      "step": 111720
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.152817341234177e-05,
+      "loss": 0.964,
+      "step": 111730
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.152741176291377e-05,
+      "loss": 1.0787,
+      "step": 111740
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1526650113485766e-05,
+      "loss": 0.9231,
+      "step": 111750
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.152588846405776e-05,
+      "loss": 1.0013,
+      "step": 111760
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.152512681462977e-05,
+      "loss": 0.9379,
+      "step": 111770
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1524365165201765e-05,
+      "loss": 1.0096,
+      "step": 111780
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.152360351577376e-05,
+      "loss": 0.9656,
+      "step": 111790
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.152284186634576e-05,
+      "loss": 1.0373,
+      "step": 111800
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.152208021691776e-05,
+      "loss": 0.9246,
+      "step": 111810
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.152131856748976e-05,
+      "loss": 0.9229,
+      "step": 111820
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.152055691806176e-05,
+      "loss": 1.0067,
+      "step": 111830
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1519795268633756e-05,
+      "loss": 0.9615,
+      "step": 111840
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1519033619205754e-05,
+      "loss": 1.0505,
+      "step": 111850
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151827196977775e-05,
+      "loss": 1.0275,
+      "step": 111860
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151751032034975e-05,
+      "loss": 0.9576,
+      "step": 111870
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151674867092175e-05,
+      "loss": 1.039,
+      "step": 111880
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1515987021493744e-05,
+      "loss": 1.0587,
+      "step": 111890
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151522537206575e-05,
+      "loss": 0.9348,
+      "step": 111900
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1514463722637745e-05,
+      "loss": 0.9319,
+      "step": 111910
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151370207320974e-05,
+      "loss": 0.8144,
+      "step": 111920
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151294042378174e-05,
+      "loss": 1.022,
+      "step": 111930
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1512178774353745e-05,
+      "loss": 0.9657,
+      "step": 111940
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151141712492574e-05,
+      "loss": 1.1437,
+      "step": 111950
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.151065547549774e-05,
+      "loss": 0.9897,
+      "step": 111960
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.150989382606974e-05,
+      "loss": 1.0598,
+      "step": 111970
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1509132176641734e-05,
+      "loss": 0.9595,
+      "step": 111980
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.150837052721374e-05,
+      "loss": 0.9997,
+      "step": 111990
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1507608877785736e-05,
+      "loss": 0.9298,
+      "step": 112000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.1506847228357734e-05,
+      "loss": 1.0743,
+      "step": 112010
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.150608557892973e-05,
+      "loss": 0.8744,
+      "step": 112020
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.1505323929501735e-05,
+      "loss": 0.9224,
+      "step": 112030
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.1504562280073726e-05,
+      "loss": 0.9164,
+      "step": 112040
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.150380063064573e-05,
+      "loss": 0.9226,
+      "step": 112050
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.150303898121773e-05,
+      "loss": 1.0939,
+      "step": 112060
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.1502277331789725e-05,
+      "loss": 1.0492,
+      "step": 112070
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.150151568236172e-05,
+      "loss": 1.0287,
+      "step": 112080
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.150075403293373e-05,
+      "loss": 0.9748,
+      "step": 112090
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149999238350572e-05,
+      "loss": 1.0425,
+      "step": 112100
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149923073407772e-05,
+      "loss": 0.9101,
+      "step": 112110
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149846908464972e-05,
+      "loss": 0.9542,
+      "step": 112120
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149770743522172e-05,
+      "loss": 1.043,
+      "step": 112130
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.1496945785793714e-05,
+      "loss": 0.9931,
+      "step": 112140
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149618413636572e-05,
+      "loss": 1.1401,
+      "step": 112150
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.1495422486937716e-05,
+      "loss": 0.9036,
+      "step": 112160
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149466083750971e-05,
+      "loss": 0.9362,
+      "step": 112170
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149389918808171e-05,
+      "loss": 0.8413,
+      "step": 112180
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149313753865371e-05,
+      "loss": 0.9843,
+      "step": 112190
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149237588922571e-05,
+      "loss": 0.8691,
+      "step": 112200
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149161423979771e-05,
+      "loss": 0.9322,
+      "step": 112210
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.149085259036971e-05,
+      "loss": 0.8665,
+      "step": 112220
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.1490090940941705e-05,
+      "loss": 1.0299,
+      "step": 112230
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.148932929151371e-05,
+      "loss": 0.9636,
+      "step": 112240
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.14885676420857e-05,
+      "loss": 0.9639,
+      "step": 112250
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1487805992657704e-05,
+      "loss": 1.0745,
+      "step": 112260
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.14870443432297e-05,
+      "loss": 0.958,
+      "step": 112270
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.14862826938017e-05,
+      "loss": 1.0764,
+      "step": 112280
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1485521044373697e-05,
+      "loss": 0.9819,
+      "step": 112290
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1484759394945694e-05,
+      "loss": 0.8873,
+      "step": 112300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.148399774551769e-05,
+      "loss": 0.9605,
+      "step": 112310
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1483236096089696e-05,
+      "loss": 0.9211,
+      "step": 112320
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.148247444666169e-05,
+      "loss": 0.9188,
+      "step": 112330
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.148171279723369e-05,
+      "loss": 1.0034,
+      "step": 112340
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.148095114780569e-05,
+      "loss": 0.9939,
+      "step": 112350
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1480189498377686e-05,
+      "loss": 0.94,
+      "step": 112360
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.147942784894969e-05,
+      "loss": 0.9098,
+      "step": 112370
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.147866619952169e-05,
+      "loss": 1.0554,
+      "step": 112380
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1477904550093685e-05,
+      "loss": 0.9423,
+      "step": 112390
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.147714290066568e-05,
+      "loss": 0.9604,
+      "step": 112400
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1476381251237686e-05,
+      "loss": 0.8801,
+      "step": 112410
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.147561960180968e-05,
+      "loss": 0.9652,
+      "step": 112420
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.147485795238168e-05,
+      "loss": 0.9456,
+      "step": 112430
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.147409630295368e-05,
+      "loss": 0.9453,
+      "step": 112440
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.1473334653525676e-05,
+      "loss": 0.8669,
+      "step": 112450
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1472573004097674e-05,
+      "loss": 0.9712,
+      "step": 112460
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.147181135466968e-05,
+      "loss": 0.9708,
+      "step": 112470
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.147104970524167e-05,
+      "loss": 0.9525,
+      "step": 112480
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.147028805581367e-05,
+      "loss": 0.9799,
+      "step": 112490
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146952640638567e-05,
+      "loss": 0.9952,
+      "step": 112500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146876475695767e-05,
+      "loss": 0.9892,
+      "step": 112510
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1468003107529665e-05,
+      "loss": 0.9707,
+      "step": 112520
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146724145810167e-05,
+      "loss": 0.8687,
+      "step": 112530
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146647980867367e-05,
+      "loss": 1.035,
+      "step": 112540
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1465718159245665e-05,
+      "loss": 0.9667,
+      "step": 112550
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146495650981766e-05,
+      "loss": 0.9153,
+      "step": 112560
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146419486038966e-05,
+      "loss": 0.9499,
+      "step": 112570
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1463433210961664e-05,
+      "loss": 0.8472,
+      "step": 112580
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146267156153366e-05,
+      "loss": 0.9408,
+      "step": 112590
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146190991210566e-05,
+      "loss": 0.9434,
+      "step": 112600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1461148262677656e-05,
+      "loss": 1.0247,
+      "step": 112610
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.146038661324966e-05,
+      "loss": 1.0331,
+      "step": 112620
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.145962496382165e-05,
+      "loss": 1.03,
+      "step": 112630
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1458863314393655e-05,
+      "loss": 0.8468,
+      "step": 112640
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.145810166496565e-05,
+      "loss": 0.9365,
+      "step": 112650
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.145734001553765e-05,
+      "loss": 1.0954,
+      "step": 112660
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.145657836610965e-05,
+      "loss": 0.9392,
+      "step": 112670
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.145581671668165e-05,
+      "loss": 0.935,
+      "step": 112680
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.145505506725364e-05,
+      "loss": 0.8601,
+      "step": 112690
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.145429341782565e-05,
+      "loss": 0.929,
+      "step": 112700
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.1453531768397644e-05,
+      "loss": 1.0759,
+      "step": 112710
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.145277011896964e-05,
+      "loss": 0.8925,
+      "step": 112720
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.145200846954164e-05,
+      "loss": 0.9629,
+      "step": 112730
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.1451246820113644e-05,
+      "loss": 1.0707,
+      "step": 112740
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.145048517068564e-05,
+      "loss": 0.9831,
+      "step": 112750
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144972352125764e-05,
+      "loss": 0.9399,
+      "step": 112760
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.1448961871829636e-05,
+      "loss": 0.9856,
+      "step": 112770
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.1448200222401633e-05,
+      "loss": 0.8522,
+      "step": 112780
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144743857297364e-05,
+      "loss": 0.908,
+      "step": 112790
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144667692354563e-05,
+      "loss": 1.0736,
+      "step": 112800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144591527411763e-05,
+      "loss": 1.0064,
+      "step": 112810
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144515362468963e-05,
+      "loss": 0.8985,
+      "step": 112820
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144439197526163e-05,
+      "loss": 1.0012,
+      "step": 112830
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.1443630325833625e-05,
+      "loss": 0.8946,
+      "step": 112840
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144286867640563e-05,
+      "loss": 1.0599,
+      "step": 112850
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144210702697762e-05,
+      "loss": 0.8822,
+      "step": 112860
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.1441345377549624e-05,
+      "loss": 0.9174,
+      "step": 112870
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.144058372812162e-05,
+      "loss": 0.9885,
+      "step": 112880
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.143982207869362e-05,
+      "loss": 1.0744,
+      "step": 112890
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1439060429265617e-05,
+      "loss": 0.9262,
+      "step": 112900
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143829877983762e-05,
+      "loss": 0.9838,
+      "step": 112910
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143753713040962e-05,
+      "loss": 1.0313,
+      "step": 112920
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1436775480981616e-05,
+      "loss": 0.9983,
+      "step": 112930
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143601383155361e-05,
+      "loss": 0.9349,
+      "step": 112940
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143525218212561e-05,
+      "loss": 1.0955,
+      "step": 112950
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1434490532697615e-05,
+      "loss": 0.8012,
+      "step": 112960
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143372888326961e-05,
+      "loss": 0.9474,
+      "step": 112970
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143296723384161e-05,
+      "loss": 0.9996,
+      "step": 112980
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143220558441361e-05,
+      "loss": 0.9326,
+      "step": 112990
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.143144393498561e-05,
+      "loss": 0.9432,
+      "step": 113000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.14306822855576e-05,
+      "loss": 1.0178,
+      "step": 113010
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1429920636129606e-05,
+      "loss": 1.0719,
+      "step": 113020
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1429158986701604e-05,
+      "loss": 0.909,
+      "step": 113030
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.14283973372736e-05,
+      "loss": 0.9566,
+      "step": 113040
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.14276356878456e-05,
+      "loss": 0.9521,
+      "step": 113050
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.14268740384176e-05,
+      "loss": 1.0131,
+      "step": 113060
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1426112388989594e-05,
+      "loss": 1.0072,
+      "step": 113070
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.14253507395616e-05,
+      "loss": 0.9076,
+      "step": 113080
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.1424589090133596e-05,
+      "loss": 0.9477,
+      "step": 113090
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.142382744070559e-05,
+      "loss": 0.9307,
+      "step": 113100
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.142306579127759e-05,
+      "loss": 0.9699,
+      "step": 113110
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1422304141849595e-05,
+      "loss": 1.0349,
+      "step": 113120
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.142154249242159e-05,
+      "loss": 0.8378,
+      "step": 113130
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.142078084299359e-05,
+      "loss": 1.0158,
+      "step": 113140
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.142001919356559e-05,
+      "loss": 0.9054,
+      "step": 113150
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1419257544137585e-05,
+      "loss": 0.9978,
+      "step": 113160
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141849589470959e-05,
+      "loss": 0.9117,
+      "step": 113170
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1417734245281586e-05,
+      "loss": 0.8879,
+      "step": 113180
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1416972595853584e-05,
+      "loss": 0.8398,
+      "step": 113190
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141621094642558e-05,
+      "loss": 1.1078,
+      "step": 113200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1415449296997585e-05,
+      "loss": 0.9209,
+      "step": 113210
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1414687647569576e-05,
+      "loss": 0.9945,
+      "step": 113220
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141392599814158e-05,
+      "loss": 0.9792,
+      "step": 113230
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141316434871357e-05,
+      "loss": 0.9012,
+      "step": 113240
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1412402699285575e-05,
+      "loss": 0.9793,
+      "step": 113250
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141164104985757e-05,
+      "loss": 0.8173,
+      "step": 113260
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141087940042957e-05,
+      "loss": 1.0038,
+      "step": 113270
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.141011775100157e-05,
+      "loss": 0.8898,
+      "step": 113280
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.140935610157357e-05,
+      "loss": 0.9769,
+      "step": 113290
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.140859445214556e-05,
+      "loss": 0.9497,
+      "step": 113300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.140783280271757e-05,
+      "loss": 0.9181,
+      "step": 113310
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1407071153289564e-05,
+      "loss": 0.9384,
+      "step": 113320
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.140630950386156e-05,
+      "loss": 0.9512,
+      "step": 113330
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1405547854433566e-05,
+      "loss": 0.9715,
+      "step": 113340
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1404786205005564e-05,
+      "loss": 0.8999,
+      "step": 113350
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.140402455557756e-05,
+      "loss": 0.9758,
+      "step": 113360
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.140326290614956e-05,
+      "loss": 0.9717,
+      "step": 113370
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.140250125672156e-05,
+      "loss": 0.9039,
+      "step": 113380
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1401739607293553e-05,
+      "loss": 0.948,
+      "step": 113390
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.140097795786556e-05,
+      "loss": 1.0249,
+      "step": 113400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1400216308437555e-05,
+      "loss": 1.0585,
+      "step": 113410
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139945465900955e-05,
+      "loss": 1.0631,
+      "step": 113420
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139869300958155e-05,
+      "loss": 1.0129,
+      "step": 113430
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1397931360153554e-05,
+      "loss": 1.0024,
+      "step": 113440
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1397169710725545e-05,
+      "loss": 0.9601,
+      "step": 113450
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139640806129755e-05,
+      "loss": 1.0067,
+      "step": 113460
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139564641186955e-05,
+      "loss": 0.9173,
+      "step": 113470
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1394884762441544e-05,
+      "loss": 0.8478,
+      "step": 113480
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139412311301354e-05,
+      "loss": 1.0471,
+      "step": 113490
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1393361463585546e-05,
+      "loss": 0.9905,
+      "step": 113500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139259981415754e-05,
+      "loss": 0.8952,
+      "step": 113510
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139183816472954e-05,
+      "loss": 1.018,
+      "step": 113520
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.139107651530154e-05,
+      "loss": 0.8601,
+      "step": 113530
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.1390314865873536e-05,
+      "loss": 0.9803,
+      "step": 113540
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.138955321644554e-05,
+      "loss": 0.8857,
+      "step": 113550
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138879156701754e-05,
+      "loss": 0.9963,
+      "step": 113560
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1388029917589535e-05,
+      "loss": 1.0213,
+      "step": 113570
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138726826816153e-05,
+      "loss": 0.9658,
+      "step": 113580
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138650661873354e-05,
+      "loss": 1.015,
+      "step": 113590
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138574496930553e-05,
+      "loss": 0.8736,
+      "step": 113600
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138498331987753e-05,
+      "loss": 0.8345,
+      "step": 113610
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138422167044953e-05,
+      "loss": 0.9951,
+      "step": 113620
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1383460021021526e-05,
+      "loss": 0.8754,
+      "step": 113630
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1382698371593524e-05,
+      "loss": 1.1529,
+      "step": 113640
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138193672216553e-05,
+      "loss": 1.0296,
+      "step": 113650
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138117507273752e-05,
+      "loss": 1.0606,
+      "step": 113660
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.138041342330952e-05,
+      "loss": 0.9014,
+      "step": 113670
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1379651773881514e-05,
+      "loss": 0.9916,
+      "step": 113680
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.137889012445352e-05,
+      "loss": 1.0847,
+      "step": 113690
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1378128475025516e-05,
+      "loss": 0.9418,
+      "step": 113700
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.137736682559751e-05,
+      "loss": 0.9536,
+      "step": 113710
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.137660517616952e-05,
+      "loss": 1.0159,
+      "step": 113720
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1375843526741515e-05,
+      "loss": 1.0161,
+      "step": 113730
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.137508187731351e-05,
+      "loss": 0.9139,
+      "step": 113740
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.137432022788551e-05,
+      "loss": 0.9109,
+      "step": 113750
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.1373558578457514e-05,
+      "loss": 0.9606,
+      "step": 113760
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1372796929029505e-05,
+      "loss": 0.9571,
+      "step": 113770
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.137203527960151e-05,
+      "loss": 0.9742,
+      "step": 113780
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1371273630173506e-05,
+      "loss": 0.9108,
+      "step": 113790
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1370511980745504e-05,
+      "loss": 0.9931,
+      "step": 113800
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.13697503313175e-05,
+      "loss": 0.9596,
+      "step": 113810
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1368988681889505e-05,
+      "loss": 0.9654,
+      "step": 113820
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1368227032461496e-05,
+      "loss": 1.0383,
+      "step": 113830
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.13674653830335e-05,
+      "loss": 1.0689,
+      "step": 113840
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.13667037336055e-05,
+      "loss": 0.9809,
+      "step": 113850
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1365942084177495e-05,
+      "loss": 0.828,
+      "step": 113860
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.136518043474949e-05,
+      "loss": 0.9767,
+      "step": 113870
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.13644187853215e-05,
+      "loss": 1.0053,
+      "step": 113880
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.136365713589349e-05,
+      "loss": 0.9546,
+      "step": 113890
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.136289548646549e-05,
+      "loss": 1.0365,
+      "step": 113900
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.136213383703749e-05,
+      "loss": 0.9002,
+      "step": 113910
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.136137218760949e-05,
+      "loss": 1.0563,
+      "step": 113920
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.136061053818149e-05,
+      "loss": 0.9594,
+      "step": 113930
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.135984888875349e-05,
+      "loss": 1.0429,
+      "step": 113940
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1359087239325486e-05,
+      "loss": 1.006,
+      "step": 113950
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.1358325589897484e-05,
+      "loss": 0.987,
+      "step": 113960
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.135756394046949e-05,
+      "loss": 0.8277,
+      "step": 113970
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.135680229104148e-05,
+      "loss": 1.0141,
+      "step": 113980
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.135604064161348e-05,
+      "loss": 0.931,
+      "step": 113990
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.135527899218548e-05,
+      "loss": 0.9283,
+      "step": 114000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.135451734275748e-05,
+      "loss": 1.0726,
+      "step": 114010
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1353755693329475e-05,
+      "loss": 0.8537,
+      "step": 114020
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.135299404390148e-05,
+      "loss": 0.8478,
+      "step": 114030
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.135223239447347e-05,
+      "loss": 1.0378,
+      "step": 114040
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1351470745045474e-05,
+      "loss": 0.9899,
+      "step": 114050
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.135070909561747e-05,
+      "loss": 0.9027,
+      "step": 114060
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134994744618947e-05,
+      "loss": 0.9988,
+      "step": 114070
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134918579676147e-05,
+      "loss": 0.8728,
+      "step": 114080
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134842414733347e-05,
+      "loss": 0.9359,
+      "step": 114090
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134766249790546e-05,
+      "loss": 0.8569,
+      "step": 114100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1346900848477466e-05,
+      "loss": 0.8783,
+      "step": 114110
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134613919904946e-05,
+      "loss": 1.0062,
+      "step": 114120
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134537754962146e-05,
+      "loss": 1.0122,
+      "step": 114130
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1344615900193465e-05,
+      "loss": 0.9377,
+      "step": 114140
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1343854250765456e-05,
+      "loss": 1.0537,
+      "step": 114150
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134309260133746e-05,
+      "loss": 1.0123,
+      "step": 114160
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134233095190946e-05,
+      "loss": 0.9476,
+      "step": 114170
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1341569302481455e-05,
+      "loss": 1.1276,
+      "step": 114180
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134080765305345e-05,
+      "loss": 1.0359,
+      "step": 114190
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.134004600362546e-05,
+      "loss": 0.9895,
+      "step": 114200
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133928435419745e-05,
+      "loss": 0.9628,
+      "step": 114210
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133852270476945e-05,
+      "loss": 0.978,
+      "step": 114220
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133776105534145e-05,
+      "loss": 1.0708,
+      "step": 114230
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.1336999405913446e-05,
+      "loss": 1.0504,
+      "step": 114240
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.1336237756485444e-05,
+      "loss": 1.0282,
+      "step": 114250
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133547610705745e-05,
+      "loss": 1.0718,
+      "step": 114260
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133471445762944e-05,
+      "loss": 0.9666,
+      "step": 114270
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133395280820144e-05,
+      "loss": 1.0322,
+      "step": 114280
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133319115877344e-05,
+      "loss": 0.9418,
+      "step": 114290
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133242950934544e-05,
+      "loss": 1.0002,
+      "step": 114300
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133166785991744e-05,
+      "loss": 0.9415,
+      "step": 114310
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133090621048944e-05,
+      "loss": 0.9113,
+      "step": 114320
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.133014456106144e-05,
+      "loss": 0.9402,
+      "step": 114330
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.1329382911633435e-05,
+      "loss": 1.058,
+      "step": 114340
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.132862126220544e-05,
+      "loss": 0.9698,
+      "step": 114350
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.132785961277743e-05,
+      "loss": 0.9072,
+      "step": 114360
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.1327097963349434e-05,
+      "loss": 0.9718,
+      "step": 114370
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.132633631392143e-05,
+      "loss": 0.8462,
+      "step": 114380
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.132557466449343e-05,
+      "loss": 0.892,
+      "step": 114390
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.1324813015065426e-05,
+      "loss": 0.8838,
+      "step": 114400
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.132405136563743e-05,
+      "loss": 0.9652,
+      "step": 114410
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.132328971620942e-05,
+      "loss": 0.9715,
+      "step": 114420
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1322528066781425e-05,
+      "loss": 0.8588,
+      "step": 114430
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.132176641735342e-05,
+      "loss": 0.9835,
+      "step": 114440
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.132100476792542e-05,
+      "loss": 0.9976,
+      "step": 114450
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.132024311849742e-05,
+      "loss": 0.9958,
+      "step": 114460
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131948146906942e-05,
+      "loss": 0.9246,
+      "step": 114470
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131871981964141e-05,
+      "loss": 0.9129,
+      "step": 114480
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131795817021342e-05,
+      "loss": 0.9412,
+      "step": 114490
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1317196520785415e-05,
+      "loss": 0.9781,
+      "step": 114500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131643487135741e-05,
+      "loss": 0.9222,
+      "step": 114510
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1315673221929416e-05,
+      "loss": 1.0113,
+      "step": 114520
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1314911572501414e-05,
+      "loss": 0.9367,
+      "step": 114530
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131414992307341e-05,
+      "loss": 1.0214,
+      "step": 114540
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131338827364541e-05,
+      "loss": 0.9335,
+      "step": 114550
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131262662421741e-05,
+      "loss": 0.9566,
+      "step": 114560
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1311864974789404e-05,
+      "loss": 0.9559,
+      "step": 114570
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.131110332536141e-05,
+      "loss": 1.0808,
+      "step": 114580
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1310341675933405e-05,
+      "loss": 1.068,
+      "step": 114590
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.13095800265054e-05,
+      "loss": 0.9768,
+      "step": 114600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.13088183770774e-05,
+      "loss": 1.0246,
+      "step": 114610
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.13080567276494e-05,
+      "loss": 1.0159,
+      "step": 114620
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.1307295078221395e-05,
+      "loss": 0.8853,
+      "step": 114630
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.13065334287934e-05,
+      "loss": 0.9861,
+      "step": 114640
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.130577177936539e-05,
+      "loss": 1.0689,
+      "step": 114650
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.1305010129937394e-05,
+      "loss": 0.9118,
+      "step": 114660
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.130424848050939e-05,
+      "loss": 1.019,
+      "step": 114670
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.130348683108139e-05,
+      "loss": 0.9765,
+      "step": 114680
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.130272518165339e-05,
+      "loss": 0.9067,
+      "step": 114690
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.130196353222539e-05,
+      "loss": 0.9871,
+      "step": 114700
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.130120188279739e-05,
+      "loss": 0.9757,
+      "step": 114710
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.1300440233369386e-05,
+      "loss": 1.0336,
+      "step": 114720
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129967858394139e-05,
+      "loss": 0.9956,
+      "step": 114730
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129891693451338e-05,
+      "loss": 0.9606,
+      "step": 114740
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.1298155285085385e-05,
+      "loss": 0.913,
+      "step": 114750
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129739363565738e-05,
+      "loss": 0.9472,
+      "step": 114760
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129663198622938e-05,
+      "loss": 1.0578,
+      "step": 114770
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129587033680138e-05,
+      "loss": 1.0355,
+      "step": 114780
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129510868737338e-05,
+      "loss": 1.0319,
+      "step": 114790
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129434703794537e-05,
+      "loss": 0.9916,
+      "step": 114800
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129358538851738e-05,
+      "loss": 0.9566,
+      "step": 114810
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.1292823739089374e-05,
+      "loss": 0.983,
+      "step": 114820
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129206208966137e-05,
+      "loss": 0.9326,
+      "step": 114830
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129130044023337e-05,
+      "loss": 1.0137,
+      "step": 114840
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.129053879080537e-05,
+      "loss": 1.0013,
+      "step": 114850
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.1289777141377364e-05,
+      "loss": 0.9074,
+      "step": 114860
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.128901549194937e-05,
+      "loss": 0.9861,
+      "step": 114870
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1288253842521366e-05,
+      "loss": 0.8234,
+      "step": 114880
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.128749219309336e-05,
+      "loss": 0.9222,
+      "step": 114890
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.128673054366536e-05,
+      "loss": 1.0765,
+      "step": 114900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1285968894237365e-05,
+      "loss": 0.9766,
+      "step": 114910
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.128520724480936e-05,
+      "loss": 0.9571,
+      "step": 114920
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.128444559538136e-05,
+      "loss": 0.8766,
+      "step": 114930
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1283683945953364e-05,
+      "loss": 0.8709,
+      "step": 114940
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1282922296525355e-05,
+      "loss": 0.9448,
+      "step": 114950
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.128216064709736e-05,
+      "loss": 1.0826,
+      "step": 114960
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1281398997669356e-05,
+      "loss": 0.9529,
+      "step": 114970
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1280637348241354e-05,
+      "loss": 0.9932,
+      "step": 114980
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127987569881335e-05,
+      "loss": 0.9054,
+      "step": 114990
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1279114049385356e-05,
+      "loss": 0.9371,
+      "step": 115000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1278352399957346e-05,
+      "loss": 1.0032,
+      "step": 115010
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127759075052935e-05,
+      "loss": 1.0782,
+      "step": 115020
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127682910110135e-05,
+      "loss": 0.9162,
+      "step": 115030
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.1276067451673345e-05,
+      "loss": 0.9415,
+      "step": 115040
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127530580224534e-05,
+      "loss": 0.9521,
+      "step": 115050
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127454415281734e-05,
+      "loss": 0.9164,
+      "step": 115060
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127378250338934e-05,
+      "loss": 0.7792,
+      "step": 115070
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.127302085396134e-05,
+      "loss": 0.9855,
+      "step": 115080
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.127225920453334e-05,
+      "loss": 1.0192,
+      "step": 115090
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.127149755510534e-05,
+      "loss": 0.9526,
+      "step": 115100
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.127073590567734e-05,
+      "loss": 0.8897,
+      "step": 115110
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126997425624933e-05,
+      "loss": 0.9896,
+      "step": 115120
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1269212606821336e-05,
+      "loss": 0.9525,
+      "step": 115130
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1268450957393334e-05,
+      "loss": 0.9768,
+      "step": 115140
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126768930796533e-05,
+      "loss": 0.9418,
+      "step": 115150
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126692765853733e-05,
+      "loss": 0.9005,
+      "step": 115160
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126616600910933e-05,
+      "loss": 0.9671,
+      "step": 115170
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1265404359681324e-05,
+      "loss": 0.9924,
+      "step": 115180
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126464271025333e-05,
+      "loss": 1.0109,
+      "step": 115190
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1263881060825325e-05,
+      "loss": 0.8969,
+      "step": 115200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126311941139732e-05,
+      "loss": 1.0575,
+      "step": 115210
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126235776196932e-05,
+      "loss": 1.0456,
+      "step": 115220
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1261596112541324e-05,
+      "loss": 0.9537,
+      "step": 115230
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1260834463113315e-05,
+      "loss": 0.9512,
+      "step": 115240
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.126007281368532e-05,
+      "loss": 1.0589,
+      "step": 115250
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.125931116425732e-05,
+      "loss": 1.0318,
+      "step": 115260
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1258549514829314e-05,
+      "loss": 1.0218,
+      "step": 115270
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.125778786540131e-05,
+      "loss": 1.0147,
+      "step": 115280
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1257026215973316e-05,
+      "loss": 0.9021,
+      "step": 115290
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.1256264566545313e-05,
+      "loss": 0.9165,
+      "step": 115300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.125550291711731e-05,
+      "loss": 0.9744,
+      "step": 115310
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.1254741267689315e-05,
+      "loss": 0.7891,
+      "step": 115320
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.1253979618261306e-05,
+      "loss": 1.0107,
+      "step": 115330
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.125321796883331e-05,
+      "loss": 0.9348,
+      "step": 115340
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.125245631940531e-05,
+      "loss": 0.93,
+      "step": 115350
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.1251694669977305e-05,
+      "loss": 0.9084,
+      "step": 115360
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.12509330205493e-05,
+      "loss": 0.8778,
+      "step": 115370
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.125017137112131e-05,
+      "loss": 0.8957,
+      "step": 115380
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.12494097216933e-05,
+      "loss": 0.9389,
+      "step": 115390
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.12486480722653e-05,
+      "loss": 0.9806,
+      "step": 115400
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.12478864228373e-05,
+      "loss": 1.0234,
+      "step": 115410
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.12471247734093e-05,
+      "loss": 0.9817,
+      "step": 115420
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.1246363123981294e-05,
+      "loss": 1.0124,
+      "step": 115430
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.12456014745533e-05,
+      "loss": 1.0409,
+      "step": 115440
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.124483982512529e-05,
+      "loss": 0.9592,
+      "step": 115450
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.124407817569729e-05,
+      "loss": 1.0348,
+      "step": 115460
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.124331652626929e-05,
+      "loss": 0.83,
+      "step": 115470
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.124255487684129e-05,
+      "loss": 0.9641,
+      "step": 115480
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.1241793227413286e-05,
+      "loss": 1.0336,
+      "step": 115490
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.124103157798529e-05,
+      "loss": 1.0377,
+      "step": 115500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.124026992855729e-05,
+      "loss": 0.977,
+      "step": 115510
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.1239508279129285e-05,
+      "loss": 0.9029,
+      "step": 115520
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.123874662970128e-05,
+      "loss": 0.8828,
+      "step": 115530
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.123798498027328e-05,
+      "loss": 0.8469,
+      "step": 115540
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1237223330845284e-05,
+      "loss": 1.0671,
+      "step": 115550
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1236461681417275e-05,
+      "loss": 1.0217,
+      "step": 115560
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.123570003198928e-05,
+      "loss": 1.0056,
+      "step": 115570
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1234938382561276e-05,
+      "loss": 0.9641,
+      "step": 115580
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1234176733133274e-05,
+      "loss": 1.0258,
+      "step": 115590
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.123341508370527e-05,
+      "loss": 1.058,
+      "step": 115600
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1232653434277276e-05,
+      "loss": 1.0635,
+      "step": 115610
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1231891784849266e-05,
+      "loss": 0.8931,
+      "step": 115620
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.123113013542127e-05,
+      "loss": 1.0518,
+      "step": 115630
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.123036848599327e-05,
+      "loss": 1.0681,
+      "step": 115640
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1229606836565265e-05,
+      "loss": 1.0423,
+      "step": 115650
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122884518713726e-05,
+      "loss": 1.1056,
+      "step": 115660
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122808353770927e-05,
+      "loss": 1.0193,
+      "step": 115670
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1227321888281265e-05,
+      "loss": 0.8495,
+      "step": 115680
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122656023885326e-05,
+      "loss": 1.0801,
+      "step": 115690
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122579858942526e-05,
+      "loss": 0.9135,
+      "step": 115700
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122503693999726e-05,
+      "loss": 0.9028,
+      "step": 115710
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122427529056926e-05,
+      "loss": 1.0142,
+      "step": 115720
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.122351364114126e-05,
+      "loss": 0.8983,
+      "step": 115730
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.1222751991713256e-05,
+      "loss": 0.9645,
+      "step": 115740
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.1221990342285254e-05,
+      "loss": 0.9672,
+      "step": 115750
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.122122869285726e-05,
+      "loss": 0.9752,
+      "step": 115760
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.122046704342925e-05,
+      "loss": 0.9332,
+      "step": 115770
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121970539400125e-05,
+      "loss": 0.9045,
+      "step": 115780
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121894374457325e-05,
+      "loss": 0.9782,
+      "step": 115790
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121818209514525e-05,
+      "loss": 1.0435,
+      "step": 115800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.1217420445717245e-05,
+      "loss": 0.9445,
+      "step": 115810
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121665879628925e-05,
+      "loss": 1.0478,
+      "step": 115820
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121589714686124e-05,
+      "loss": 0.9651,
+      "step": 115830
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.1215135497433244e-05,
+      "loss": 0.9505,
+      "step": 115840
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121437384800524e-05,
+      "loss": 0.8772,
+      "step": 115850
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121361219857724e-05,
+      "loss": 0.9657,
+      "step": 115860
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121285054914924e-05,
+      "loss": 1.0257,
+      "step": 115870
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121208889972124e-05,
+      "loss": 1.0348,
+      "step": 115880
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.121132725029324e-05,
+      "loss": 0.9986,
+      "step": 115890
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.1210565600865236e-05,
+      "loss": 0.9591,
+      "step": 115900
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.120980395143724e-05,
+      "loss": 0.9373,
+      "step": 115910
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.120904230200923e-05,
+      "loss": 0.9512,
+      "step": 115920
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.1208280652581235e-05,
+      "loss": 0.9304,
+      "step": 115930
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.120751900315323e-05,
+      "loss": 0.8928,
+      "step": 115940
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.120675735372523e-05,
+      "loss": 1.008,
+      "step": 115950
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120599570429723e-05,
+      "loss": 0.9861,
+      "step": 115960
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120523405486923e-05,
+      "loss": 0.9065,
+      "step": 115970
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120447240544122e-05,
+      "loss": 0.8206,
+      "step": 115980
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120371075601323e-05,
+      "loss": 0.9217,
+      "step": 115990
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120294910658522e-05,
+      "loss": 0.8911,
+      "step": 116000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120218745715722e-05,
+      "loss": 0.9725,
+      "step": 116010
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120142580772922e-05,
+      "loss": 1.0379,
+      "step": 116020
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.120066415830122e-05,
+      "loss": 1.0344,
+      "step": 116030
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.1199902508873214e-05,
+      "loss": 0.971,
+      "step": 116040
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119914085944522e-05,
+      "loss": 0.971,
+      "step": 116050
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.1198379210017216e-05,
+      "loss": 0.8931,
+      "step": 116060
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119761756058921e-05,
+      "loss": 0.9207,
+      "step": 116070
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119685591116121e-05,
+      "loss": 1.09,
+      "step": 116080
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119609426173321e-05,
+      "loss": 1.0161,
+      "step": 116090
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119533261230521e-05,
+      "loss": 0.9594,
+      "step": 116100
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119457096287721e-05,
+      "loss": 0.9226,
+      "step": 116110
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119380931344921e-05,
+      "loss": 0.9066,
+      "step": 116120
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.1193047664021205e-05,
+      "loss": 0.9617,
+      "step": 116130
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.119228601459321e-05,
+      "loss": 1.0169,
+      "step": 116140
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.11915243651652e-05,
+      "loss": 0.9331,
+      "step": 116150
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.1190762715737204e-05,
+      "loss": 1.1098,
+      "step": 116160
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.11900010663092e-05,
+      "loss": 0.912,
+      "step": 116170
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.11892394168812e-05,
+      "loss": 1.0847,
+      "step": 116180
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1188477767453196e-05,
+      "loss": 0.9735,
+      "step": 116190
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.11877161180252e-05,
+      "loss": 1.1393,
+      "step": 116200
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118695446859719e-05,
+      "loss": 1.0685,
+      "step": 116210
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1186192819169196e-05,
+      "loss": 1.0398,
+      "step": 116220
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118543116974119e-05,
+      "loss": 0.9909,
+      "step": 116230
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118466952031319e-05,
+      "loss": 0.971,
+      "step": 116240
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118390787088519e-05,
+      "loss": 0.947,
+      "step": 116250
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118314622145719e-05,
+      "loss": 1.0185,
+      "step": 116260
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118238457202919e-05,
+      "loss": 0.9471,
+      "step": 116270
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118162292260119e-05,
+      "loss": 0.9015,
+      "step": 116280
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1180861273173185e-05,
+      "loss": 1.0815,
+      "step": 116290
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.118009962374518e-05,
+      "loss": 0.8622,
+      "step": 116300
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1179337974317186e-05,
+      "loss": 0.9498,
+      "step": 116310
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1178576324889184e-05,
+      "loss": 1.0518,
+      "step": 116320
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.117781467546118e-05,
+      "loss": 0.9461,
+      "step": 116330
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.117705302603318e-05,
+      "loss": 0.9942,
+      "step": 116340
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.117629137660518e-05,
+      "loss": 0.8443,
+      "step": 116350
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1175529727177174e-05,
+      "loss": 0.8649,
+      "step": 116360
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.117476807774918e-05,
+      "loss": 0.915,
+      "step": 116370
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.1174006428321175e-05,
+      "loss": 0.8629,
+      "step": 116380
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.117324477889317e-05,
+      "loss": 0.8901,
+      "step": 116390
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.117248312946517e-05,
+      "loss": 0.9016,
+      "step": 116400
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1171721480037175e-05,
+      "loss": 0.8928,
+      "step": 116410
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1170959830609165e-05,
+      "loss": 1.0494,
+      "step": 116420
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.117019818118117e-05,
+      "loss": 0.8947,
+      "step": 116430
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116943653175317e-05,
+      "loss": 0.8507,
+      "step": 116440
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1168674882325164e-05,
+      "loss": 0.8972,
+      "step": 116450
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116791323289716e-05,
+      "loss": 1.1211,
+      "step": 116460
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116715158346916e-05,
+      "loss": 0.9947,
+      "step": 116470
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1166389934041164e-05,
+      "loss": 0.9388,
+      "step": 116480
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116562828461316e-05,
+      "loss": 0.8892,
+      "step": 116490
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116486663518516e-05,
+      "loss": 1.0544,
+      "step": 116500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1164104985757156e-05,
+      "loss": 0.9178,
+      "step": 116510
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116334333632916e-05,
+      "loss": 0.902,
+      "step": 116520
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116258168690115e-05,
+      "loss": 0.9016,
+      "step": 116530
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1161820037473155e-05,
+      "loss": 1.0302,
+      "step": 116540
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116105838804515e-05,
+      "loss": 1.0106,
+      "step": 116550
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.116029673861715e-05,
+      "loss": 1.0427,
+      "step": 116560
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.115953508918915e-05,
+      "loss": 0.9557,
+      "step": 116570
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.115877343976115e-05,
+      "loss": 0.9718,
+      "step": 116580
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.115801179033314e-05,
+      "loss": 0.9722,
+      "step": 116590
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.115725014090515e-05,
+      "loss": 1.107,
+      "step": 116600
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.1156488491477144e-05,
+      "loss": 1.0394,
+      "step": 116610
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.115572684204914e-05,
+      "loss": 0.8289,
+      "step": 116620
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.115496519262114e-05,
+      "loss": 0.9676,
+      "step": 116630
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1154203543193143e-05,
+      "loss": 0.97,
+      "step": 116640
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.115344189376514e-05,
+      "loss": 0.9618,
+      "step": 116650
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.115268024433714e-05,
+      "loss": 0.8052,
+      "step": 116660
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1151918594909136e-05,
+      "loss": 0.881,
+      "step": 116670
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.115115694548113e-05,
+      "loss": 1.0629,
+      "step": 116680
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.115039529605314e-05,
+      "loss": 1.0248,
+      "step": 116690
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1149633646625135e-05,
+      "loss": 0.9231,
+      "step": 116700
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.114887199719713e-05,
+      "loss": 0.9907,
+      "step": 116710
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.114811034776913e-05,
+      "loss": 1.0793,
+      "step": 116720
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1147348698341134e-05,
+      "loss": 1.1132,
+      "step": 116730
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1146587048913125e-05,
+      "loss": 1.002,
+      "step": 116740
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.114582539948513e-05,
+      "loss": 0.9728,
+      "step": 116750
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1145063750057127e-05,
+      "loss": 0.9279,
+      "step": 116760
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1144302100629124e-05,
+      "loss": 0.9394,
+      "step": 116770
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.114354045120112e-05,
+      "loss": 0.9381,
+      "step": 116780
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1142778801773126e-05,
+      "loss": 0.9332,
+      "step": 116790
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1142017152345116e-05,
+      "loss": 1.0163,
+      "step": 116800
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.114125550291712e-05,
+      "loss": 0.9714,
+      "step": 116810
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.114049385348912e-05,
+      "loss": 0.9102,
+      "step": 116820
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.1139732204061116e-05,
+      "loss": 0.9574,
+      "step": 116830
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.113897055463311e-05,
+      "loss": 0.9378,
+      "step": 116840
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.113820890520512e-05,
+      "loss": 1.0117,
+      "step": 116850
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.1137447255777115e-05,
+      "loss": 1.0116,
+      "step": 116860
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.113668560634911e-05,
+      "loss": 1.0723,
+      "step": 116870
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.113592395692111e-05,
+      "loss": 0.9661,
+      "step": 116880
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.113516230749311e-05,
+      "loss": 0.9857,
+      "step": 116890
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.113440065806511e-05,
+      "loss": 1.0357,
+      "step": 116900
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.11336390086371e-05,
+      "loss": 0.8984,
+      "step": 116910
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.1132877359209106e-05,
+      "loss": 0.9464,
+      "step": 116920
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.1132115709781104e-05,
+      "loss": 0.9016,
+      "step": 116930
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.11313540603531e-05,
+      "loss": 1.0173,
+      "step": 116940
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.11305924109251e-05,
+      "loss": 0.9705,
+      "step": 116950
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.11298307614971e-05,
+      "loss": 1.1046,
+      "step": 116960
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.1129069112069094e-05,
+      "loss": 0.926,
+      "step": 116970
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.11283074626411e-05,
+      "loss": 0.9531,
+      "step": 116980
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.1127545813213095e-05,
+      "loss": 1.0057,
+      "step": 116990
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.112678416378509e-05,
+      "loss": 1.0683,
+      "step": 117000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.112602251435709e-05,
+      "loss": 1.0086,
+      "step": 117010
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.1125260864929095e-05,
+      "loss": 0.9109,
+      "step": 117020
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.112449921550109e-05,
+      "loss": 1.0332,
+      "step": 117030
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.112373756607309e-05,
+      "loss": 0.929,
+      "step": 117040
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.112297591664509e-05,
+      "loss": 0.9335,
+      "step": 117050
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1122214267217084e-05,
+      "loss": 1.0425,
+      "step": 117060
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.112145261778909e-05,
+      "loss": 0.8734,
+      "step": 117070
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1120690968361086e-05,
+      "loss": 0.9348,
+      "step": 117080
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1119929318933084e-05,
+      "loss": 1.0237,
+      "step": 117090
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111916766950508e-05,
+      "loss": 0.8949,
+      "step": 117100
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1118406020077085e-05,
+      "loss": 0.8218,
+      "step": 117110
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1117644370649076e-05,
+      "loss": 1.0257,
+      "step": 117120
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111688272122108e-05,
+      "loss": 0.9993,
+      "step": 117130
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111612107179308e-05,
+      "loss": 0.9912,
+      "step": 117140
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1115359422365075e-05,
+      "loss": 0.9899,
+      "step": 117150
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111459777293707e-05,
+      "loss": 1.0096,
+      "step": 117160
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111383612350908e-05,
+      "loss": 1.0294,
+      "step": 117170
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111307447408107e-05,
+      "loss": 1.0332,
+      "step": 117180
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111231282465307e-05,
+      "loss": 0.9398,
+      "step": 117190
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111155117522507e-05,
+      "loss": 0.9524,
+      "step": 117200
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.111078952579707e-05,
+      "loss": 0.9641,
+      "step": 117210
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1110027876369064e-05,
+      "loss": 1.0112,
+      "step": 117220
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.110926622694107e-05,
+      "loss": 0.9946,
+      "step": 117230
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1108504577513066e-05,
+      "loss": 1.0849,
+      "step": 117240
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.1107742928085063e-05,
+      "loss": 0.9624,
+      "step": 117250
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.110698127865706e-05,
+      "loss": 1.0098,
+      "step": 117260
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.110621962922906e-05,
+      "loss": 0.9697,
+      "step": 117270
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.110545797980106e-05,
+      "loss": 0.8817,
+      "step": 117280
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.110469633037306e-05,
+      "loss": 0.925,
+      "step": 117290
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.110393468094506e-05,
+      "loss": 1.1424,
+      "step": 117300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1103173031517055e-05,
+      "loss": 1.0273,
+      "step": 117310
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.110241138208906e-05,
+      "loss": 0.9799,
+      "step": 117320
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.110164973266105e-05,
+      "loss": 0.9426,
+      "step": 117330
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1100888083233054e-05,
+      "loss": 0.9596,
+      "step": 117340
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.110012643380505e-05,
+      "loss": 0.9325,
+      "step": 117350
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109936478437705e-05,
+      "loss": 0.9012,
+      "step": 117360
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1098603134949047e-05,
+      "loss": 0.9671,
+      "step": 117370
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1097841485521044e-05,
+      "loss": 0.9547,
+      "step": 117380
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109707983609304e-05,
+      "loss": 0.9129,
+      "step": 117390
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1096318186665046e-05,
+      "loss": 0.9133,
+      "step": 117400
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109555653723704e-05,
+      "loss": 0.9661,
+      "step": 117410
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109479488780904e-05,
+      "loss": 0.9141,
+      "step": 117420
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109403323838104e-05,
+      "loss": 1.0392,
+      "step": 117430
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1093271588953036e-05,
+      "loss": 0.9661,
+      "step": 117440
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109250993952504e-05,
+      "loss": 0.9872,
+      "step": 117450
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109174829009704e-05,
+      "loss": 0.8749,
+      "step": 117460
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1090986640669035e-05,
+      "loss": 0.9318,
+      "step": 117470
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.109022499124103e-05,
+      "loss": 1.0282,
+      "step": 117480
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.1089463341813037e-05,
+      "loss": 0.922,
+      "step": 117490
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108870169238503e-05,
+      "loss": 1.0042,
+      "step": 117500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108794004295703e-05,
+      "loss": 0.9932,
+      "step": 117510
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108717839352903e-05,
+      "loss": 1.0729,
+      "step": 117520
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1086416744101026e-05,
+      "loss": 0.9844,
+      "step": 117530
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1085655094673024e-05,
+      "loss": 0.9892,
+      "step": 117540
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108489344524503e-05,
+      "loss": 1.022,
+      "step": 117550
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108413179581702e-05,
+      "loss": 0.9714,
+      "step": 117560
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108337014638902e-05,
+      "loss": 0.9718,
+      "step": 117570
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108260849696102e-05,
+      "loss": 0.8808,
+      "step": 117580
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108184684753302e-05,
+      "loss": 1.0111,
+      "step": 117590
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1081085198105015e-05,
+      "loss": 0.9983,
+      "step": 117600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.108032354867702e-05,
+      "loss": 0.9536,
+      "step": 117610
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.107956189924902e-05,
+      "loss": 0.8723,
+      "step": 117620
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1078800249821015e-05,
+      "loss": 0.987,
+      "step": 117630
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.107803860039301e-05,
+      "loss": 1.0149,
+      "step": 117640
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.107727695096501e-05,
+      "loss": 0.9327,
+      "step": 117650
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1076515301537014e-05,
+      "loss": 0.8744,
+      "step": 117660
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.107575365210901e-05,
+      "loss": 1.0442,
+      "step": 117670
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.107499200268101e-05,
+      "loss": 1.1011,
+      "step": 117680
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1074230353253006e-05,
+      "loss": 0.881,
+      "step": 117690
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.107346870382501e-05,
+      "loss": 0.9964,
+      "step": 117700
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.1072707054397e-05,
+      "loss": 0.9446,
+      "step": 117710
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1071945404969005e-05,
+      "loss": 1.0467,
+      "step": 117720
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1071183755541e-05,
+      "loss": 1.0189,
+      "step": 117730
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1070422106113e-05,
+      "loss": 0.9807,
+      "step": 117740
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1069660456685e-05,
+      "loss": 0.9072,
+      "step": 117750
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1068898807257e-05,
+      "loss": 0.9968,
+      "step": 117760
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.106813715782899e-05,
+      "loss": 0.9123,
+      "step": 117770
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1067375508401e-05,
+      "loss": 1.1258,
+      "step": 117780
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1066613858972994e-05,
+      "loss": 1.069,
+      "step": 117790
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.106585220954499e-05,
+      "loss": 0.8722,
+      "step": 117800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.106509056011699e-05,
+      "loss": 0.9704,
+      "step": 117810
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1064328910688994e-05,
+      "loss": 0.9424,
+      "step": 117820
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.106356726126099e-05,
+      "loss": 1.0654,
+      "step": 117830
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.106280561183299e-05,
+      "loss": 0.919,
+      "step": 117840
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1062043962404986e-05,
+      "loss": 0.9695,
+      "step": 117850
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1061282312976983e-05,
+      "loss": 1.1794,
+      "step": 117860
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.106052066354899e-05,
+      "loss": 1.0254,
+      "step": 117870
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.105975901412098e-05,
+      "loss": 1.0299,
+      "step": 117880
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.105899736469298e-05,
+      "loss": 0.9088,
+      "step": 117890
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.105823571526498e-05,
+      "loss": 0.9561,
+      "step": 117900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.105747406583698e-05,
+      "loss": 0.9828,
+      "step": 117910
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.1056712416408975e-05,
+      "loss": 1.0206,
+      "step": 117920
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.105595076698098e-05,
+      "loss": 0.9923,
+      "step": 117930
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.105518911755297e-05,
+      "loss": 0.874,
+      "step": 117940
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.1054427468124974e-05,
+      "loss": 0.9604,
+      "step": 117950
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.105366581869697e-05,
+      "loss": 0.9128,
+      "step": 117960
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.105290416926897e-05,
+      "loss": 0.9534,
+      "step": 117970
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.105214251984097e-05,
+      "loss": 0.9262,
+      "step": 117980
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.105138087041297e-05,
+      "loss": 0.9526,
+      "step": 117990
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.105061922098496e-05,
+      "loss": 0.9387,
+      "step": 118000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.1049857571556966e-05,
+      "loss": 1.0175,
+      "step": 118010
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104909592212896e-05,
+      "loss": 0.8276,
+      "step": 118020
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104833427270096e-05,
+      "loss": 0.915,
+      "step": 118030
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.1047572623272965e-05,
+      "loss": 1.0373,
+      "step": 118040
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104681097384496e-05,
+      "loss": 0.894,
+      "step": 118050
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104604932441696e-05,
+      "loss": 1.033,
+      "step": 118060
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104528767498896e-05,
+      "loss": 0.9471,
+      "step": 118070
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104452602556096e-05,
+      "loss": 0.9977,
+      "step": 118080
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104376437613295e-05,
+      "loss": 0.9474,
+      "step": 118090
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.1043002726704957e-05,
+      "loss": 0.9072,
+      "step": 118100
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.1042241077276954e-05,
+      "loss": 0.9325,
+      "step": 118110
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104147942784895e-05,
+      "loss": 0.8743,
+      "step": 118120
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.104071777842095e-05,
+      "loss": 1.0651,
+      "step": 118130
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.103995612899295e-05,
+      "loss": 1.0455,
+      "step": 118140
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1039194479564944e-05,
+      "loss": 0.9638,
+      "step": 118150
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103843283013695e-05,
+      "loss": 0.8946,
+      "step": 118160
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1037671180708946e-05,
+      "loss": 0.9517,
+      "step": 118170
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103690953128094e-05,
+      "loss": 0.9585,
+      "step": 118180
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103614788185294e-05,
+      "loss": 1.1595,
+      "step": 118190
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1035386232424945e-05,
+      "loss": 0.9671,
+      "step": 118200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103462458299694e-05,
+      "loss": 1.018,
+      "step": 118210
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103386293356894e-05,
+      "loss": 0.98,
+      "step": 118220
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103310128414094e-05,
+      "loss": 0.9748,
+      "step": 118230
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1032339634712935e-05,
+      "loss": 0.9086,
+      "step": 118240
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.103157798528494e-05,
+      "loss": 0.9346,
+      "step": 118250
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1030816335856936e-05,
+      "loss": 0.9559,
+      "step": 118260
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1030054686428934e-05,
+      "loss": 0.9861,
+      "step": 118270
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102929303700093e-05,
+      "loss": 0.9795,
+      "step": 118280
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102853138757293e-05,
+      "loss": 0.9566,
+      "step": 118290
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1027769738144926e-05,
+      "loss": 1.0195,
+      "step": 118300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102700808871693e-05,
+      "loss": 0.9663,
+      "step": 118310
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102624643928892e-05,
+      "loss": 0.8578,
+      "step": 118320
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.1025484789860925e-05,
+      "loss": 0.8514,
+      "step": 118330
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102472314043292e-05,
+      "loss": 0.985,
+      "step": 118340
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102396149100492e-05,
+      "loss": 0.9685,
+      "step": 118350
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.102319984157692e-05,
+      "loss": 0.9109,
+      "step": 118360
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.102243819214892e-05,
+      "loss": 0.8725,
+      "step": 118370
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.102167654272091e-05,
+      "loss": 0.9979,
+      "step": 118380
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.102091489329292e-05,
+      "loss": 0.9684,
+      "step": 118390
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1020153243864914e-05,
+      "loss": 0.96,
+      "step": 118400
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.101939159443691e-05,
+      "loss": 0.9333,
+      "step": 118410
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1018629945008916e-05,
+      "loss": 1.1462,
+      "step": 118420
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1017868295580914e-05,
+      "loss": 0.9605,
+      "step": 118430
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.101710664615291e-05,
+      "loss": 0.9251,
+      "step": 118440
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.101634499672491e-05,
+      "loss": 1.0322,
+      "step": 118450
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.101558334729691e-05,
+      "loss": 0.9372,
+      "step": 118460
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1014821697868903e-05,
+      "loss": 0.9297,
+      "step": 118470
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.101406004844091e-05,
+      "loss": 1.1274,
+      "step": 118480
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1013298399012905e-05,
+      "loss": 0.9474,
+      "step": 118490
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.10125367495849e-05,
+      "loss": 1.0054,
+      "step": 118500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.10117751001569e-05,
+      "loss": 1.0794,
+      "step": 118510
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1011013450728904e-05,
+      "loss": 0.9343,
+      "step": 118520
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1010251801300895e-05,
+      "loss": 0.8901,
+      "step": 118530
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.10094901518729e-05,
+      "loss": 0.9645,
+      "step": 118540
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.10087285024449e-05,
+      "loss": 0.9377,
+      "step": 118550
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1007966853016894e-05,
+      "loss": 1.0296,
+      "step": 118560
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.100720520358889e-05,
+      "loss": 0.9567,
+      "step": 118570
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.1006443554160896e-05,
+      "loss": 1.0426,
+      "step": 118580
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.100568190473289e-05,
+      "loss": 1.0854,
+      "step": 118590
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.100492025530489e-05,
+      "loss": 0.9326,
+      "step": 118600
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.100415860587689e-05,
+      "loss": 1.1242,
+      "step": 118610
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.1003396956448886e-05,
+      "loss": 0.89,
+      "step": 118620
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.100263530702089e-05,
+      "loss": 1.0075,
+      "step": 118630
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.100187365759289e-05,
+      "loss": 0.9982,
+      "step": 118640
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.1001112008164885e-05,
+      "loss": 1.0055,
+      "step": 118650
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.100035035873688e-05,
+      "loss": 0.8711,
+      "step": 118660
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099958870930889e-05,
+      "loss": 0.9503,
+      "step": 118670
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099882705988088e-05,
+      "loss": 1.0271,
+      "step": 118680
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099806541045288e-05,
+      "loss": 1.0066,
+      "step": 118690
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099730376102488e-05,
+      "loss": 0.9114,
+      "step": 118700
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.0996542111596877e-05,
+      "loss": 0.9857,
+      "step": 118710
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.0995780462168874e-05,
+      "loss": 0.8859,
+      "step": 118720
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099501881274088e-05,
+      "loss": 0.9181,
+      "step": 118730
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099425716331287e-05,
+      "loss": 1.0595,
+      "step": 118740
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099349551388487e-05,
+      "loss": 0.8551,
+      "step": 118750
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.0992733864456864e-05,
+      "loss": 0.9803,
+      "step": 118760
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099197221502887e-05,
+      "loss": 1.012,
+      "step": 118770
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.0991210565600866e-05,
+      "loss": 0.9936,
+      "step": 118780
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.099044891617286e-05,
+      "loss": 1.0802,
+      "step": 118790
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.098968726674486e-05,
+      "loss": 0.9874,
+      "step": 118800
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0988925617316865e-05,
+      "loss": 1.0253,
+      "step": 118810
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.098816396788886e-05,
+      "loss": 0.989,
+      "step": 118820
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.098740231846086e-05,
+      "loss": 1.0107,
+      "step": 118830
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0986640669032864e-05,
+      "loss": 1.0085,
+      "step": 118840
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0985879019604855e-05,
+      "loss": 1.0464,
+      "step": 118850
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.098511737017686e-05,
+      "loss": 0.9396,
+      "step": 118860
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0984355720748856e-05,
+      "loss": 0.9587,
+      "step": 118870
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0983594071320854e-05,
+      "loss": 0.9812,
+      "step": 118880
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.098283242189285e-05,
+      "loss": 1.0266,
+      "step": 118890
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0982070772464856e-05,
+      "loss": 1.0845,
+      "step": 118900
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0981309123036846e-05,
+      "loss": 1.0184,
+      "step": 118910
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.098054747360885e-05,
+      "loss": 1.0438,
+      "step": 118920
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097978582418085e-05,
+      "loss": 0.8807,
+      "step": 118930
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.0979024174752845e-05,
+      "loss": 0.9674,
+      "step": 118940
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097826252532484e-05,
+      "loss": 0.9818,
+      "step": 118950
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097750087589685e-05,
+      "loss": 1.0107,
+      "step": 118960
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097673922646884e-05,
+      "loss": 0.9519,
+      "step": 118970
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097597757704084e-05,
+      "loss": 1.0051,
+      "step": 118980
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097521592761284e-05,
+      "loss": 1.0022,
+      "step": 118990
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097445427818484e-05,
+      "loss": 0.9145,
+      "step": 119000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097369262875684e-05,
+      "loss": 1.0103,
+      "step": 119010
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.097293097932884e-05,
+      "loss": 1.0486,
+      "step": 119020
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0972169329900836e-05,
+      "loss": 0.9706,
+      "step": 119030
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0971407680472834e-05,
+      "loss": 0.9196,
+      "step": 119040
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.097064603104484e-05,
+      "loss": 0.9578,
+      "step": 119050
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096988438161683e-05,
+      "loss": 1.0219,
+      "step": 119060
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096912273218883e-05,
+      "loss": 1.0507,
+      "step": 119070
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096836108276083e-05,
+      "loss": 0.9487,
+      "step": 119080
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096759943333283e-05,
+      "loss": 0.9425,
+      "step": 119090
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0966837783904825e-05,
+      "loss": 0.8803,
+      "step": 119100
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096607613447683e-05,
+      "loss": 0.911,
+      "step": 119110
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096531448504882e-05,
+      "loss": 1.0182,
+      "step": 119120
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0964552835620824e-05,
+      "loss": 1.1507,
+      "step": 119130
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096379118619282e-05,
+      "loss": 0.9781,
+      "step": 119140
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096302953676482e-05,
+      "loss": 0.9529,
+      "step": 119150
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096226788733682e-05,
+      "loss": 0.9073,
+      "step": 119160
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096150623790882e-05,
+      "loss": 0.953,
+      "step": 119170
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.096074458848081e-05,
+      "loss": 0.908,
+      "step": 119180
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0959982939052816e-05,
+      "loss": 1.0522,
+      "step": 119190
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0959221289624813e-05,
+      "loss": 0.963,
+      "step": 119200
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.095845964019681e-05,
+      "loss": 0.8964,
+      "step": 119210
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0957697990768815e-05,
+      "loss": 0.9948,
+      "step": 119220
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0956936341340806e-05,
+      "loss": 1.002,
+      "step": 119230
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.095617469191281e-05,
+      "loss": 0.9588,
+      "step": 119240
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.095541304248481e-05,
+      "loss": 0.9129,
+      "step": 119250
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.0954651393056805e-05,
+      "loss": 1.0404,
+      "step": 119260
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.09538897436288e-05,
+      "loss": 1.0698,
+      "step": 119270
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.095312809420081e-05,
+      "loss": 0.8674,
+      "step": 119280
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.09523664447728e-05,
+      "loss": 0.8989,
+      "step": 119290
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.09516047953448e-05,
+      "loss": 0.9288,
+      "step": 119300
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.09508431459168e-05,
+      "loss": 0.9284,
+      "step": 119310
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.0950081496488797e-05,
+      "loss": 0.8936,
+      "step": 119320
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.0949319847060794e-05,
+      "loss": 0.964,
+      "step": 119330
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.09485581976328e-05,
+      "loss": 1.0336,
+      "step": 119340
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094779654820479e-05,
+      "loss": 1.0571,
+      "step": 119350
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094703489877679e-05,
+      "loss": 1.0396,
+      "step": 119360
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094627324934879e-05,
+      "loss": 1.0158,
+      "step": 119370
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094551159992079e-05,
+      "loss": 0.9622,
+      "step": 119380
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.0944749950492786e-05,
+      "loss": 0.991,
+      "step": 119390
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094398830106479e-05,
+      "loss": 0.9648,
+      "step": 119400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094322665163679e-05,
+      "loss": 0.9015,
+      "step": 119410
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.0942465002208785e-05,
+      "loss": 0.9587,
+      "step": 119420
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094170335278079e-05,
+      "loss": 1.0939,
+      "step": 119430
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.094094170335278e-05,
+      "loss": 0.9547,
+      "step": 119440
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.0940180053924784e-05,
+      "loss": 1.0114,
+      "step": 119450
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.093941840449678e-05,
+      "loss": 0.9432,
+      "step": 119460
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093865675506878e-05,
+      "loss": 0.9757,
+      "step": 119470
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0937895105640776e-05,
+      "loss": 1.0353,
+      "step": 119480
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093713345621278e-05,
+      "loss": 1.0528,
+      "step": 119490
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093637180678477e-05,
+      "loss": 0.9755,
+      "step": 119500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0935610157356776e-05,
+      "loss": 1.0756,
+      "step": 119510
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093484850792877e-05,
+      "loss": 0.9687,
+      "step": 119520
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093408685850077e-05,
+      "loss": 0.9507,
+      "step": 119530
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093332520907277e-05,
+      "loss": 0.9507,
+      "step": 119540
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093256355964477e-05,
+      "loss": 1.1313,
+      "step": 119550
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093180191021676e-05,
+      "loss": 0.8505,
+      "step": 119560
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.093104026078877e-05,
+      "loss": 1.0496,
+      "step": 119570
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0930278611360765e-05,
+      "loss": 0.9507,
+      "step": 119580
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092951696193276e-05,
+      "loss": 0.9437,
+      "step": 119590
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0928755312504766e-05,
+      "loss": 1.0094,
+      "step": 119600
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0927993663076764e-05,
+      "loss": 1.0722,
+      "step": 119610
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092723201364876e-05,
+      "loss": 0.8723,
+      "step": 119620
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092647036422076e-05,
+      "loss": 0.9423,
+      "step": 119630
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092570871479276e-05,
+      "loss": 0.9711,
+      "step": 119640
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.0924947065364754e-05,
+      "loss": 1.0255,
+      "step": 119650
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092418541593676e-05,
+      "loss": 0.9526,
+      "step": 119660
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092342376650875e-05,
+      "loss": 0.9823,
+      "step": 119670
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.092266211708075e-05,
+      "loss": 1.0137,
+      "step": 119680
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.092190046765275e-05,
+      "loss": 1.037,
+      "step": 119690
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.092113881822475e-05,
+      "loss": 0.8945,
+      "step": 119700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.0920377168796745e-05,
+      "loss": 0.989,
+      "step": 119710
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091961551936875e-05,
+      "loss": 0.953,
+      "step": 119720
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091885386994074e-05,
+      "loss": 0.9838,
+      "step": 119730
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.0918092220512744e-05,
+      "loss": 0.973,
+      "step": 119740
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091733057108474e-05,
+      "loss": 0.9883,
+      "step": 119750
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091656892165674e-05,
+      "loss": 0.9192,
+      "step": 119760
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091580727222874e-05,
+      "loss": 0.9805,
+      "step": 119770
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091504562280074e-05,
+      "loss": 1.0329,
+      "step": 119780
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091428397337274e-05,
+      "loss": 0.9721,
+      "step": 119790
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.0913522323944736e-05,
+      "loss": 0.9977,
+      "step": 119800
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091276067451674e-05,
+      "loss": 0.9794,
+      "step": 119810
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091199902508873e-05,
+      "loss": 1.0016,
+      "step": 119820
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.0911237375660735e-05,
+      "loss": 0.9967,
+      "step": 119830
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.091047572623273e-05,
+      "loss": 0.9158,
+      "step": 119840
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.090971407680473e-05,
+      "loss": 0.9823,
+      "step": 119850
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.090895242737673e-05,
+      "loss": 1.0162,
+      "step": 119860
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.090819077794873e-05,
+      "loss": 1.0241,
+      "step": 119870
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.090742912852072e-05,
+      "loss": 1.0426,
+      "step": 119880
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.090666747909273e-05,
+      "loss": 0.9173,
+      "step": 119890
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.0905905829664724e-05,
+      "loss": 0.9736,
+      "step": 119900
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.090514418023672e-05,
+      "loss": 0.9254,
+      "step": 119910
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.090438253080872e-05,
+      "loss": 1.0066,
+      "step": 119920
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.090362088138072e-05,
+      "loss": 0.9625,
+      "step": 119930
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0902859231952714e-05,
+      "loss": 1.0732,
+      "step": 119940
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.090209758252472e-05,
+      "loss": 0.96,
+      "step": 119950
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0901335933096716e-05,
+      "loss": 0.9392,
+      "step": 119960
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.090057428366871e-05,
+      "loss": 0.9986,
+      "step": 119970
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.089981263424071e-05,
+      "loss": 1.0094,
+      "step": 119980
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0899050984812715e-05,
+      "loss": 0.8991,
+      "step": 119990
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.089828933538471e-05,
+      "loss": 1.0948,
+      "step": 120000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.089752768595671e-05,
+      "loss": 0.8515,
+      "step": 120010
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0896766036528714e-05,
+      "loss": 0.9439,
+      "step": 120020
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0896004387100705e-05,
+      "loss": 0.9902,
+      "step": 120030
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.089524273767271e-05,
+      "loss": 0.952,
+      "step": 120040
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0894481088244707e-05,
+      "loss": 0.924,
+      "step": 120050
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0893719438816704e-05,
+      "loss": 0.912,
+      "step": 120060
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.08929577893887e-05,
+      "loss": 1.0269,
+      "step": 120070
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0892196139960706e-05,
+      "loss": 0.9653,
+      "step": 120080
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0891434490532696e-05,
+      "loss": 1.0302,
+      "step": 120090
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.08906728411047e-05,
+      "loss": 1.0268,
+      "step": 120100
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.08899111916767e-05,
+      "loss": 0.8907,
+      "step": 120110
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0889149542248696e-05,
+      "loss": 0.9791,
+      "step": 120120
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088838789282069e-05,
+      "loss": 0.8959,
+      "step": 120130
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088762624339269e-05,
+      "loss": 1.0028,
+      "step": 120140
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088686459396469e-05,
+      "loss": 1.062,
+      "step": 120150
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088610294453669e-05,
+      "loss": 0.9472,
+      "step": 120160
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088534129510869e-05,
+      "loss": 0.9935,
+      "step": 120170
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088457964568069e-05,
+      "loss": 0.9471,
+      "step": 120180
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0883817996252685e-05,
+      "loss": 0.9509,
+      "step": 120190
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088305634682468e-05,
+      "loss": 0.8634,
+      "step": 120200
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0882294697396686e-05,
+      "loss": 0.9878,
+      "step": 120210
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0881533047968684e-05,
+      "loss": 0.9887,
+      "step": 120220
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088077139854068e-05,
+      "loss": 0.9741,
+      "step": 120230
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.088000974911268e-05,
+      "loss": 0.9858,
+      "step": 120240
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.087924809968468e-05,
+      "loss": 1.029,
+      "step": 120250
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0878486450256674e-05,
+      "loss": 0.9169,
+      "step": 120260
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.087772480082868e-05,
+      "loss": 0.9883,
+      "step": 120270
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0876963151400675e-05,
+      "loss": 1.0272,
+      "step": 120280
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.087620150197267e-05,
+      "loss": 0.9327,
+      "step": 120290
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.087543985254467e-05,
+      "loss": 0.8699,
+      "step": 120300
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0874678203116675e-05,
+      "loss": 0.9373,
+      "step": 120310
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.0873916553688665e-05,
+      "loss": 0.9387,
+      "step": 120320
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.087315490426067e-05,
+      "loss": 0.9297,
+      "step": 120330
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.087239325483267e-05,
+      "loss": 0.9535,
+      "step": 120340
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0871631605404664e-05,
+      "loss": 0.8905,
+      "step": 120350
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.087086995597666e-05,
+      "loss": 1.0406,
+      "step": 120360
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0870108306548666e-05,
+      "loss": 1.1319,
+      "step": 120370
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0869346657120664e-05,
+      "loss": 1.0119,
+      "step": 120380
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086858500769266e-05,
+      "loss": 1.0006,
+      "step": 120390
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0867823358264665e-05,
+      "loss": 1.005,
+      "step": 120400
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0867061708836656e-05,
+      "loss": 1.0328,
+      "step": 120410
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086630005940866e-05,
+      "loss": 0.861,
+      "step": 120420
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086553840998066e-05,
+      "loss": 0.9423,
+      "step": 120430
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0864776760552655e-05,
+      "loss": 0.9442,
+      "step": 120440
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086401511112465e-05,
+      "loss": 0.8959,
+      "step": 120450
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086325346169666e-05,
+      "loss": 0.9656,
+      "step": 120460
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086249181226865e-05,
+      "loss": 0.9488,
+      "step": 120470
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086173016284065e-05,
+      "loss": 1.1729,
+      "step": 120480
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086096851341265e-05,
+      "loss": 1.0104,
+      "step": 120490
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.086020686398465e-05,
+      "loss": 0.9151,
+      "step": 120500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.0859445214556644e-05,
+      "loss": 1.1571,
+      "step": 120510
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.085868356512865e-05,
+      "loss": 0.8972,
+      "step": 120520
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.085792191570064e-05,
+      "loss": 0.8805,
+      "step": 120530
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.085716026627264e-05,
+      "loss": 0.9534,
+      "step": 120540
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.085639861684464e-05,
+      "loss": 0.9723,
+      "step": 120550
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.085563696741664e-05,
+      "loss": 1.0552,
+      "step": 120560
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0854875317988636e-05,
+      "loss": 0.9827,
+      "step": 120570
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.085411366856064e-05,
+      "loss": 0.9228,
+      "step": 120580
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.085335201913264e-05,
+      "loss": 0.9538,
+      "step": 120590
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0852590369704635e-05,
+      "loss": 0.9481,
+      "step": 120600
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.085182872027663e-05,
+      "loss": 0.9569,
+      "step": 120610
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.085106707084863e-05,
+      "loss": 0.9615,
+      "step": 120620
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0850305421420634e-05,
+      "loss": 0.9329,
+      "step": 120630
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0849543771992625e-05,
+      "loss": 1.0111,
+      "step": 120640
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.084878212256463e-05,
+      "loss": 0.8285,
+      "step": 120650
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0848020473136627e-05,
+      "loss": 0.9401,
+      "step": 120660
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0847258823708624e-05,
+      "loss": 0.836,
+      "step": 120670
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.084649717428062e-05,
+      "loss": 0.8611,
+      "step": 120680
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0845735524852626e-05,
+      "loss": 0.9278,
+      "step": 120690
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0844973875424616e-05,
+      "loss": 0.8899,
+      "step": 120700
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.084421222599662e-05,
+      "loss": 0.8759,
+      "step": 120710
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.084345057656862e-05,
+      "loss": 0.915,
+      "step": 120720
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0842688927140616e-05,
+      "loss": 0.936,
+      "step": 120730
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.084192727771261e-05,
+      "loss": 1.0406,
+      "step": 120740
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.084116562828462e-05,
+      "loss": 1.0074,
+      "step": 120750
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0840403978856615e-05,
+      "loss": 0.9172,
+      "step": 120760
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.083964232942861e-05,
+      "loss": 0.9722,
+      "step": 120770
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.083888068000061e-05,
+      "loss": 0.9623,
+      "step": 120780
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.083811903057261e-05,
+      "loss": 1.0947,
+      "step": 120790
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.083735738114461e-05,
+      "loss": 0.9576,
+      "step": 120800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.083659573171661e-05,
+      "loss": 0.9883,
+      "step": 120810
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0835834082288606e-05,
+      "loss": 0.9489,
+      "step": 120820
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0835072432860604e-05,
+      "loss": 0.9653,
+      "step": 120830
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.083431078343261e-05,
+      "loss": 0.949,
+      "step": 120840
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.08335491340046e-05,
+      "loss": 0.9152,
+      "step": 120850
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.08327874845766e-05,
+      "loss": 1.0933,
+      "step": 120860
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.08320258351486e-05,
+      "loss": 0.9716,
+      "step": 120870
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.08312641857206e-05,
+      "loss": 0.9639,
+      "step": 120880
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0830502536292595e-05,
+      "loss": 1.0473,
+      "step": 120890
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.08297408868646e-05,
+      "loss": 0.886,
+      "step": 120900
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.082897923743659e-05,
+      "loss": 1.0157,
+      "step": 120910
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0828217588008595e-05,
+      "loss": 1.0075,
+      "step": 120920
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.082745593858059e-05,
+      "loss": 0.9426,
+      "step": 120930
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.082669428915259e-05,
+      "loss": 0.9894,
+      "step": 120940
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.082593263972459e-05,
+      "loss": 0.8983,
+      "step": 120950
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.082517099029659e-05,
+      "loss": 0.9954,
+      "step": 120960
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.082440934086859e-05,
+      "loss": 0.9633,
+      "step": 120970
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0823647691440586e-05,
+      "loss": 0.9832,
+      "step": 120980
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0822886042012584e-05,
+      "loss": 0.9823,
+      "step": 120990
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.082212439258458e-05,
+      "loss": 1.0107,
+      "step": 121000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.0821362743156585e-05,
+      "loss": 0.9854,
+      "step": 121010
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.082060109372858e-05,
+      "loss": 0.8927,
+      "step": 121020
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081983944430058e-05,
+      "loss": 0.9868,
+      "step": 121030
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081907779487258e-05,
+      "loss": 0.9112,
+      "step": 121040
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081831614544458e-05,
+      "loss": 0.9051,
+      "step": 121050
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081755449601657e-05,
+      "loss": 1.0325,
+      "step": 121060
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081679284658858e-05,
+      "loss": 0.9164,
+      "step": 121070
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081603119716057e-05,
+      "loss": 0.9297,
+      "step": 121080
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081526954773257e-05,
+      "loss": 1.0099,
+      "step": 121090
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081450789830457e-05,
+      "loss": 0.909,
+      "step": 121100
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081374624887657e-05,
+      "loss": 0.9051,
+      "step": 121110
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.0812984599448564e-05,
+      "loss": 0.9822,
+      "step": 121120
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081222295002057e-05,
+      "loss": 0.8539,
+      "step": 121130
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.0811461300592566e-05,
+      "loss": 0.8048,
+      "step": 121140
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.081069965116456e-05,
+      "loss": 0.9466,
+      "step": 121150
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.080993800173656e-05,
+      "loss": 1.0067,
+      "step": 121160
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.080917635230856e-05,
+      "loss": 0.8776,
+      "step": 121170
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.080841470288056e-05,
+      "loss": 0.9967,
+      "step": 121180
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.080765305345256e-05,
+      "loss": 0.9836,
+      "step": 121190
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.080689140402456e-05,
+      "loss": 0.9953,
+      "step": 121200
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.0806129754596555e-05,
+      "loss": 0.9906,
+      "step": 121210
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.080536810516856e-05,
+      "loss": 0.9537,
+      "step": 121220
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.080460645574055e-05,
+      "loss": 0.988,
+      "step": 121230
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.0803844806312554e-05,
+      "loss": 0.9425,
+      "step": 121240
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.080308315688455e-05,
+      "loss": 1.0006,
+      "step": 121250
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.080232150745655e-05,
+      "loss": 1.0139,
+      "step": 121260
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.0801559858028547e-05,
+      "loss": 0.9763,
+      "step": 121270
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.080079820860055e-05,
+      "loss": 0.9789,
+      "step": 121280
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.080003655917254e-05,
+      "loss": 0.9902,
+      "step": 121290
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.0799274909744546e-05,
+      "loss": 1.098,
+      "step": 121300
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079851326031654e-05,
+      "loss": 0.9563,
+      "step": 121310
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079775161088854e-05,
+      "loss": 0.9893,
+      "step": 121320
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079698996146054e-05,
+      "loss": 0.924,
+      "step": 121330
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079622831203254e-05,
+      "loss": 0.9305,
+      "step": 121340
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079546666260454e-05,
+      "loss": 0.8694,
+      "step": 121350
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079470501317654e-05,
+      "loss": 1.1009,
+      "step": 121360
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.0793943363748535e-05,
+      "loss": 0.9759,
+      "step": 121370
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079318171432053e-05,
+      "loss": 0.8786,
+      "step": 121380
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.0792420064892536e-05,
+      "loss": 1.0209,
+      "step": 121390
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.0791658415464534e-05,
+      "loss": 1.0637,
+      "step": 121400
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079089676603653e-05,
+      "loss": 0.967,
+      "step": 121410
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.079013511660853e-05,
+      "loss": 0.8965,
+      "step": 121420
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.078937346718053e-05,
+      "loss": 0.9258,
+      "step": 121430
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0788611817752524e-05,
+      "loss": 0.8543,
+      "step": 121440
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078785016832453e-05,
+      "loss": 0.9899,
+      "step": 121450
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0787088518896526e-05,
+      "loss": 1.0445,
+      "step": 121460
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078632686946852e-05,
+      "loss": 0.9213,
+      "step": 121470
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078556522004052e-05,
+      "loss": 1.0096,
+      "step": 121480
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0784803570612525e-05,
+      "loss": 0.9498,
+      "step": 121490
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0784041921184515e-05,
+      "loss": 0.9031,
+      "step": 121500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078328027175652e-05,
+      "loss": 0.956,
+      "step": 121510
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078251862232852e-05,
+      "loss": 0.8683,
+      "step": 121520
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0781756972900515e-05,
+      "loss": 1.0465,
+      "step": 121530
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078099532347251e-05,
+      "loss": 1.112,
+      "step": 121540
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.078023367404451e-05,
+      "loss": 1.0846,
+      "step": 121550
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0779472024616514e-05,
+      "loss": 0.9637,
+      "step": 121560
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.077871037518851e-05,
+      "loss": 1.0855,
+      "step": 121570
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.077794872576051e-05,
+      "loss": 0.952,
+      "step": 121580
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0777187076332506e-05,
+      "loss": 1.0632,
+      "step": 121590
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.077642542690451e-05,
+      "loss": 0.9622,
+      "step": 121600
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.07756637774765e-05,
+      "loss": 1.0202,
+      "step": 121610
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0774902128048505e-05,
+      "loss": 0.8209,
+      "step": 121620
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.07741404786205e-05,
+      "loss": 0.8831,
+      "step": 121630
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.07733788291925e-05,
+      "loss": 1.1075,
+      "step": 121640
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.07726171797645e-05,
+      "loss": 0.8957,
+      "step": 121650
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.07718555303365e-05,
+      "loss": 1.0438,
+      "step": 121660
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.077109388090849e-05,
+      "loss": 0.9613,
+      "step": 121670
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.07703322314805e-05,
+      "loss": 0.9287,
+      "step": 121680
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0769570582052494e-05,
+      "loss": 0.9544,
+      "step": 121690
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076880893262449e-05,
+      "loss": 0.9763,
+      "step": 121700
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076804728319649e-05,
+      "loss": 0.9109,
+      "step": 121710
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0767285633768494e-05,
+      "loss": 0.881,
+      "step": 121720
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076652398434049e-05,
+      "loss": 0.904,
+      "step": 121730
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076576233491249e-05,
+      "loss": 0.9172,
+      "step": 121740
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0765000685484486e-05,
+      "loss": 0.9139,
+      "step": 121750
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0764239036056483e-05,
+      "loss": 1.0022,
+      "step": 121760
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076347738662849e-05,
+      "loss": 1.0076,
+      "step": 121770
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0762715737200485e-05,
+      "loss": 1.0267,
+      "step": 121780
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076195408777248e-05,
+      "loss": 0.9101,
+      "step": 121790
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.076119243834448e-05,
+      "loss": 0.9125,
+      "step": 121800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0760430788916484e-05,
+      "loss": 0.9436,
+      "step": 121810
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0759669139488475e-05,
+      "loss": 0.9915,
+      "step": 121820
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.075890749006048e-05,
+      "loss": 0.9254,
+      "step": 121830
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.075814584063248e-05,
+      "loss": 0.9733,
+      "step": 121840
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0757384191204474e-05,
+      "loss": 1.0411,
+      "step": 121850
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.075662254177647e-05,
+      "loss": 0.9085,
+      "step": 121860
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.0755860892348476e-05,
+      "loss": 0.9866,
+      "step": 121870
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0755099242920467e-05,
+      "loss": 1.0576,
+      "step": 121880
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.075433759349247e-05,
+      "loss": 0.9321,
+      "step": 121890
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.075357594406447e-05,
+      "loss": 0.9817,
+      "step": 121900
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0752814294636466e-05,
+      "loss": 0.9543,
+      "step": 121910
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.075205264520846e-05,
+      "loss": 1.0583,
+      "step": 121920
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.075129099578047e-05,
+      "loss": 1.1606,
+      "step": 121930
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0750529346352465e-05,
+      "loss": 0.9185,
+      "step": 121940
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074976769692446e-05,
+      "loss": 1.0374,
+      "step": 121950
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074900604749646e-05,
+      "loss": 1.0947,
+      "step": 121960
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074824439806846e-05,
+      "loss": 1.0151,
+      "step": 121970
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074748274864046e-05,
+      "loss": 0.9741,
+      "step": 121980
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074672109921245e-05,
+      "loss": 0.9896,
+      "step": 121990
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0745959449784456e-05,
+      "loss": 0.9607,
+      "step": 122000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0745197800356454e-05,
+      "loss": 0.9612,
+      "step": 122010
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074443615092845e-05,
+      "loss": 0.9097,
+      "step": 122020
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074367450150045e-05,
+      "loss": 0.8889,
+      "step": 122030
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074291285207245e-05,
+      "loss": 0.9146,
+      "step": 122040
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0742151202644444e-05,
+      "loss": 0.9574,
+      "step": 122050
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.074138955321645e-05,
+      "loss": 0.9601,
+      "step": 122060
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.0740627903788446e-05,
+      "loss": 0.9408,
+      "step": 122070
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.073986625436044e-05,
+      "loss": 1.1002,
+      "step": 122080
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.073910460493244e-05,
+      "loss": 0.9421,
+      "step": 122090
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0738342955504445e-05,
+      "loss": 0.9986,
+      "step": 122100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.073758130607644e-05,
+      "loss": 0.9894,
+      "step": 122110
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.073681965664844e-05,
+      "loss": 0.9011,
+      "step": 122120
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.073605800722044e-05,
+      "loss": 1.059,
+      "step": 122130
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0735296357792435e-05,
+      "loss": 1.035,
+      "step": 122140
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.073453470836444e-05,
+      "loss": 0.9559,
+      "step": 122150
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0733773058936436e-05,
+      "loss": 0.9621,
+      "step": 122160
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0733011409508434e-05,
+      "loss": 0.9021,
+      "step": 122170
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.073224976008043e-05,
+      "loss": 1.1102,
+      "step": 122180
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0731488110652435e-05,
+      "loss": 0.9724,
+      "step": 122190
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0730726461224426e-05,
+      "loss": 1.0756,
+      "step": 122200
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072996481179643e-05,
+      "loss": 0.8915,
+      "step": 122210
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072920316236843e-05,
+      "loss": 0.9811,
+      "step": 122220
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0728441512940425e-05,
+      "loss": 0.9081,
+      "step": 122230
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072767986351242e-05,
+      "loss": 0.8992,
+      "step": 122240
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072691821408443e-05,
+      "loss": 1.0469,
+      "step": 122250
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072615656465642e-05,
+      "loss": 0.8901,
+      "step": 122260
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072539491522842e-05,
+      "loss": 1.0433,
+      "step": 122270
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072463326580042e-05,
+      "loss": 1.0472,
+      "step": 122280
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072387161637242e-05,
+      "loss": 0.9766,
+      "step": 122290
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.0723109966944414e-05,
+      "loss": 1.0666,
+      "step": 122300
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.072234831751642e-05,
+      "loss": 1.0694,
+      "step": 122310
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0721586668088416e-05,
+      "loss": 1.0436,
+      "step": 122320
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0720825018660414e-05,
+      "loss": 0.8566,
+      "step": 122330
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.072006336923241e-05,
+      "loss": 0.9857,
+      "step": 122340
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.071930171980441e-05,
+      "loss": 0.9646,
+      "step": 122350
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.071854007037641e-05,
+      "loss": 0.9595,
+      "step": 122360
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.071777842094841e-05,
+      "loss": 0.9922,
+      "step": 122370
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.071701677152041e-05,
+      "loss": 0.9674,
+      "step": 122380
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0716255122092405e-05,
+      "loss": 1.0367,
+      "step": 122390
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.071549347266441e-05,
+      "loss": 0.8833,
+      "step": 122400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.07147318232364e-05,
+      "loss": 0.977,
+      "step": 122410
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0713970173808404e-05,
+      "loss": 0.986,
+      "step": 122420
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.07132085243804e-05,
+      "loss": 0.8876,
+      "step": 122430
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.07124468749524e-05,
+      "loss": 0.979,
+      "step": 122440
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.07116852255244e-05,
+      "loss": 1.025,
+      "step": 122450
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0710923576096394e-05,
+      "loss": 0.8552,
+      "step": 122460
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.071016192666839e-05,
+      "loss": 0.888,
+      "step": 122470
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0709400277240396e-05,
+      "loss": 1.0468,
+      "step": 122480
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0708638627812387e-05,
+      "loss": 1.0191,
+      "step": 122490
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.070787697838439e-05,
+      "loss": 0.9899,
+      "step": 122500
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.070711532895639e-05,
+      "loss": 0.9127,
+      "step": 122510
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0706353679528386e-05,
+      "loss": 1.088,
+      "step": 122520
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070559203010039e-05,
+      "loss": 0.8855,
+      "step": 122530
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070483038067239e-05,
+      "loss": 1.0045,
+      "step": 122540
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.0704068731244385e-05,
+      "loss": 1.0016,
+      "step": 122550
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070330708181638e-05,
+      "loss": 1.0921,
+      "step": 122560
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070254543238839e-05,
+      "loss": 0.9541,
+      "step": 122570
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070178378296038e-05,
+      "loss": 1.0312,
+      "step": 122580
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070102213353238e-05,
+      "loss": 0.9327,
+      "step": 122590
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.070026048410438e-05,
+      "loss": 1.0202,
+      "step": 122600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.0699498834676376e-05,
+      "loss": 0.9123,
+      "step": 122610
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.0698737185248374e-05,
+      "loss": 0.8826,
+      "step": 122620
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069797553582038e-05,
+      "loss": 0.9176,
+      "step": 122630
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069721388639237e-05,
+      "loss": 0.8678,
+      "step": 122640
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069645223696437e-05,
+      "loss": 0.9687,
+      "step": 122650
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069569058753637e-05,
+      "loss": 1.0669,
+      "step": 122660
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069492893810837e-05,
+      "loss": 1.0372,
+      "step": 122670
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.0694167288680366e-05,
+      "loss": 0.9098,
+      "step": 122680
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069340563925237e-05,
+      "loss": 0.959,
+      "step": 122690
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069264398982437e-05,
+      "loss": 0.9096,
+      "step": 122700
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.0691882340396365e-05,
+      "loss": 0.9526,
+      "step": 122710
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069112069096836e-05,
+      "loss": 0.9947,
+      "step": 122720
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.069035904154036e-05,
+      "loss": 0.9962,
+      "step": 122730
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.0689597392112364e-05,
+      "loss": 0.9833,
+      "step": 122740
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068883574268436e-05,
+      "loss": 0.9106,
+      "step": 122750
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068807409325636e-05,
+      "loss": 0.9526,
+      "step": 122760
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.0687312443828356e-05,
+      "loss": 1.0033,
+      "step": 122770
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068655079440036e-05,
+      "loss": 0.9756,
+      "step": 122780
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068578914497235e-05,
+      "loss": 1.0508,
+      "step": 122790
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.0685027495544355e-05,
+      "loss": 1.0197,
+      "step": 122800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068426584611635e-05,
+      "loss": 1.1067,
+      "step": 122810
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068350419668835e-05,
+      "loss": 0.8925,
+      "step": 122820
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068274254726035e-05,
+      "loss": 0.9365,
+      "step": 122830
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068198089783235e-05,
+      "loss": 1.1478,
+      "step": 122840
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068121924840434e-05,
+      "loss": 0.9219,
+      "step": 122850
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.068045759897635e-05,
+      "loss": 1.0347,
+      "step": 122860
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.0679695949548345e-05,
+      "loss": 0.9619,
+      "step": 122870
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067893430012034e-05,
+      "loss": 0.9447,
+      "step": 122880
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067817265069234e-05,
+      "loss": 0.9694,
+      "step": 122890
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067741100126434e-05,
+      "loss": 1.0784,
+      "step": 122900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067664935183634e-05,
+      "loss": 0.9869,
+      "step": 122910
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067588770240834e-05,
+      "loss": 1.005,
+      "step": 122920
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.0675126052980336e-05,
+      "loss": 0.9275,
+      "step": 122930
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.0674364403552334e-05,
+      "loss": 0.9189,
+      "step": 122940
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067360275412434e-05,
+      "loss": 0.874,
+      "step": 122950
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.067284110469633e-05,
+      "loss": 0.8959,
+      "step": 122960
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.067207945526833e-05,
+      "loss": 0.9897,
+      "step": 122970
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.067131780584033e-05,
+      "loss": 1.1138,
+      "step": 122980
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.067055615641233e-05,
+      "loss": 0.9729,
+      "step": 122990
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.0669794506984325e-05,
+      "loss": 1.0469,
+      "step": 123000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066903285755633e-05,
+      "loss": 0.9212,
+      "step": 123010
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066827120812832e-05,
+      "loss": 1.0853,
+      "step": 123020
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.0667509558700324e-05,
+      "loss": 1.1657,
+      "step": 123030
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066674790927232e-05,
+      "loss": 0.9824,
+      "step": 123040
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066598625984432e-05,
+      "loss": 0.9739,
+      "step": 123050
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066522461041632e-05,
+      "loss": 0.9872,
+      "step": 123060
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066446296098832e-05,
+      "loss": 0.9015,
+      "step": 123070
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066370131156031e-05,
+      "loss": 0.9205,
+      "step": 123080
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.0662939662132316e-05,
+      "loss": 0.9736,
+      "step": 123090
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066217801270431e-05,
+      "loss": 0.9594,
+      "step": 123100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.066141636327631e-05,
+      "loss": 0.9629,
+      "step": 123110
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.0660654713848315e-05,
+      "loss": 1.0134,
+      "step": 123120
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.065989306442031e-05,
+      "loss": 0.9011,
+      "step": 123130
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.065913141499231e-05,
+      "loss": 0.9443,
+      "step": 123140
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.065836976556431e-05,
+      "loss": 0.8474,
+      "step": 123150
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.065760811613631e-05,
+      "loss": 0.9979,
+      "step": 123160
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.06568464667083e-05,
+      "loss": 0.9126,
+      "step": 123170
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.065608481728031e-05,
+      "loss": 0.9769,
+      "step": 123180
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0655323167852304e-05,
+      "loss": 0.995,
+      "step": 123190
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.06545615184243e-05,
+      "loss": 0.9165,
+      "step": 123200
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.06537998689963e-05,
+      "loss": 0.8777,
+      "step": 123210
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.06530382195683e-05,
+      "loss": 0.9023,
+      "step": 123220
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0652276570140294e-05,
+      "loss": 0.9097,
+      "step": 123230
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.06515149207123e-05,
+      "loss": 0.9735,
+      "step": 123240
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0650753271284296e-05,
+      "loss": 0.9665,
+      "step": 123250
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064999162185629e-05,
+      "loss": 1.0208,
+      "step": 123260
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064922997242829e-05,
+      "loss": 0.8824,
+      "step": 123270
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0648468323000295e-05,
+      "loss": 1.0098,
+      "step": 123280
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0647706673572286e-05,
+      "loss": 0.9597,
+      "step": 123290
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064694502414429e-05,
+      "loss": 0.9179,
+      "step": 123300
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064618337471629e-05,
+      "loss": 0.9831,
+      "step": 123310
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0645421725288285e-05,
+      "loss": 0.977,
+      "step": 123320
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064466007586029e-05,
+      "loss": 0.9675,
+      "step": 123330
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0643898426432286e-05,
+      "loss": 0.9412,
+      "step": 123340
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0643136777004284e-05,
+      "loss": 1.0701,
+      "step": 123350
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064237512757628e-05,
+      "loss": 1.0399,
+      "step": 123360
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064161347814828e-05,
+      "loss": 1.0986,
+      "step": 123370
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.0640851828720276e-05,
+      "loss": 1.0619,
+      "step": 123380
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.064009017929228e-05,
+      "loss": 1.0386,
+      "step": 123390
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.063932852986427e-05,
+      "loss": 1.0346,
+      "step": 123400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0638566880436275e-05,
+      "loss": 0.9409,
+      "step": 123410
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063780523100827e-05,
+      "loss": 0.9985,
+      "step": 123420
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063704358158027e-05,
+      "loss": 1.0832,
+      "step": 123430
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063628193215227e-05,
+      "loss": 1.0121,
+      "step": 123440
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063552028272427e-05,
+      "loss": 0.9869,
+      "step": 123450
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063475863329626e-05,
+      "loss": 0.9953,
+      "step": 123460
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063399698386827e-05,
+      "loss": 0.9235,
+      "step": 123470
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0633235334440265e-05,
+      "loss": 1.0649,
+      "step": 123480
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063247368501226e-05,
+      "loss": 0.9726,
+      "step": 123490
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0631712035584266e-05,
+      "loss": 0.969,
+      "step": 123500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0630950386156264e-05,
+      "loss": 1.0432,
+      "step": 123510
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.063018873672826e-05,
+      "loss": 0.9818,
+      "step": 123520
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.062942708730026e-05,
+      "loss": 1.0132,
+      "step": 123530
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.062866543787226e-05,
+      "loss": 0.8956,
+      "step": 123540
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0627903788444254e-05,
+      "loss": 0.9974,
+      "step": 123550
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.062714213901626e-05,
+      "loss": 1.0138,
+      "step": 123560
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0626380489588255e-05,
+      "loss": 0.9488,
+      "step": 123570
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.062561884016025e-05,
+      "loss": 1.0791,
+      "step": 123580
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.062485719073225e-05,
+      "loss": 1.0353,
+      "step": 123590
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0624095541304254e-05,
+      "loss": 0.9365,
+      "step": 123600
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.0623333891876245e-05,
+      "loss": 1.0951,
+      "step": 123610
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.062257224244825e-05,
+      "loss": 0.9389,
+      "step": 123620
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.062181059302025e-05,
+      "loss": 0.954,
+      "step": 123630
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.0621048943592244e-05,
+      "loss": 0.9912,
+      "step": 123640
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.062028729416424e-05,
+      "loss": 1.024,
+      "step": 123650
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.0619525644736246e-05,
+      "loss": 1.0276,
+      "step": 123660
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061876399530824e-05,
+      "loss": 0.9227,
+      "step": 123670
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061800234588024e-05,
+      "loss": 0.9485,
+      "step": 123680
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061724069645224e-05,
+      "loss": 1.0253,
+      "step": 123690
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.0616479047024236e-05,
+      "loss": 1.0137,
+      "step": 123700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061571739759624e-05,
+      "loss": 0.9346,
+      "step": 123710
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061495574816824e-05,
+      "loss": 0.9548,
+      "step": 123720
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.0614194098740235e-05,
+      "loss": 0.8917,
+      "step": 123730
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061343244931223e-05,
+      "loss": 0.9742,
+      "step": 123740
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061267079988424e-05,
+      "loss": 0.9568,
+      "step": 123750
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061190915045623e-05,
+      "loss": 0.8645,
+      "step": 123760
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061114750102823e-05,
+      "loss": 1.0028,
+      "step": 123770
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.061038585160023e-05,
+      "loss": 0.8803,
+      "step": 123780
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.060962420217223e-05,
+      "loss": 1.0695,
+      "step": 123790
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.0608862552744224e-05,
+      "loss": 0.9577,
+      "step": 123800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.060810090331623e-05,
+      "loss": 0.9657,
+      "step": 123810
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.060733925388822e-05,
+      "loss": 0.9898,
+      "step": 123820
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.060657760446022e-05,
+      "loss": 0.91,
+      "step": 123830
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.0605815955032214e-05,
+      "loss": 1.0016,
+      "step": 123840
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.060505430560422e-05,
+      "loss": 1.1055,
+      "step": 123850
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0604292656176216e-05,
+      "loss": 1.0205,
+      "step": 123860
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.060353100674821e-05,
+      "loss": 1.0579,
+      "step": 123870
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.060276935732021e-05,
+      "loss": 0.9028,
+      "step": 123880
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0602007707892215e-05,
+      "loss": 1.077,
+      "step": 123890
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.060124605846421e-05,
+      "loss": 0.9744,
+      "step": 123900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.060048440903621e-05,
+      "loss": 1.0709,
+      "step": 123910
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0599722759608214e-05,
+      "loss": 1.0319,
+      "step": 123920
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0598961110180205e-05,
+      "loss": 0.9652,
+      "step": 123930
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.059819946075221e-05,
+      "loss": 0.9087,
+      "step": 123940
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0597437811324206e-05,
+      "loss": 1.0075,
+      "step": 123950
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0596676161896204e-05,
+      "loss": 1.1079,
+      "step": 123960
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.05959145124682e-05,
+      "loss": 1.0016,
+      "step": 123970
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0595152863040206e-05,
+      "loss": 1.0125,
+      "step": 123980
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0594391213612196e-05,
+      "loss": 1.0084,
+      "step": 123990
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.05936295641842e-05,
+      "loss": 1.0006,
+      "step": 124000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.05928679147562e-05,
+      "loss": 0.9416,
+      "step": 124010
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.0592106265328195e-05,
+      "loss": 1.1689,
+      "step": 124020
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.059134461590019e-05,
+      "loss": 0.9341,
+      "step": 124030
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.05905829664722e-05,
+      "loss": 0.9324,
+      "step": 124040
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.058982131704419e-05,
+      "loss": 0.9637,
+      "step": 124050
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.058905966761619e-05,
+      "loss": 0.8922,
+      "step": 124060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058829801818819e-05,
+      "loss": 0.9586,
+      "step": 124070
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058753636876019e-05,
+      "loss": 1.0744,
+      "step": 124080
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0586774719332185e-05,
+      "loss": 1.05,
+      "step": 124090
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058601306990419e-05,
+      "loss": 1.0201,
+      "step": 124100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0585251420476186e-05,
+      "loss": 0.9271,
+      "step": 124110
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0584489771048184e-05,
+      "loss": 1.1215,
+      "step": 124120
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058372812162019e-05,
+      "loss": 0.9194,
+      "step": 124130
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058296647219218e-05,
+      "loss": 0.8473,
+      "step": 124140
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058220482276418e-05,
+      "loss": 0.8899,
+      "step": 124150
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058144317333618e-05,
+      "loss": 1.108,
+      "step": 124160
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.058068152390818e-05,
+      "loss": 0.9667,
+      "step": 124170
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0579919874480175e-05,
+      "loss": 1.0098,
+      "step": 124180
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057915822505218e-05,
+      "loss": 0.9881,
+      "step": 124190
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057839657562417e-05,
+      "loss": 1.0959,
+      "step": 124200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0577634926196174e-05,
+      "loss": 0.9144,
+      "step": 124210
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057687327676817e-05,
+      "loss": 0.9437,
+      "step": 124220
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057611162734017e-05,
+      "loss": 0.8707,
+      "step": 124230
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057534997791217e-05,
+      "loss": 0.8964,
+      "step": 124240
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057458832848417e-05,
+      "loss": 0.8874,
+      "step": 124250
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.057382667905616e-05,
+      "loss": 0.9105,
+      "step": 124260
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0573065029628166e-05,
+      "loss": 1.0282,
+      "step": 124270
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.0572303380200163e-05,
+      "loss": 1.0765,
+      "step": 124280
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.057154173077216e-05,
+      "loss": 0.9523,
+      "step": 124290
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.0570780081344165e-05,
+      "loss": 1.0621,
+      "step": 124300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.0570018431916156e-05,
+      "loss": 0.9351,
+      "step": 124310
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056925678248816e-05,
+      "loss": 0.8305,
+      "step": 124320
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056849513306016e-05,
+      "loss": 0.929,
+      "step": 124330
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.0567733483632155e-05,
+      "loss": 0.9367,
+      "step": 124340
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056697183420415e-05,
+      "loss": 0.9736,
+      "step": 124350
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056621018477616e-05,
+      "loss": 0.8777,
+      "step": 124360
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056544853534815e-05,
+      "loss": 1.0295,
+      "step": 124370
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056468688592015e-05,
+      "loss": 0.8867,
+      "step": 124380
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056392523649215e-05,
+      "loss": 1.0422,
+      "step": 124390
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056316358706415e-05,
+      "loss": 1.0842,
+      "step": 124400
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.0562401937636144e-05,
+      "loss": 0.9783,
+      "step": 124410
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056164028820815e-05,
+      "loss": 0.8831,
+      "step": 124420
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056087863878014e-05,
+      "loss": 1.074,
+      "step": 124430
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.056011698935214e-05,
+      "loss": 0.883,
+      "step": 124440
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.055935533992414e-05,
+      "loss": 1.0363,
+      "step": 124450
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.055859369049614e-05,
+      "loss": 0.9349,
+      "step": 124460
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.0557832041068136e-05,
+      "loss": 0.987,
+      "step": 124470
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.055707039164014e-05,
+      "loss": 0.9039,
+      "step": 124480
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.055630874221214e-05,
+      "loss": 0.9868,
+      "step": 124490
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.0555547092784135e-05,
+      "loss": 0.9893,
+      "step": 124500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.055478544335614e-05,
+      "loss": 0.9724,
+      "step": 124510
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.055402379392813e-05,
+      "loss": 1.0055,
+      "step": 124520
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.0553262144500134e-05,
+      "loss": 0.9723,
+      "step": 124530
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.055250049507213e-05,
+      "loss": 0.9053,
+      "step": 124540
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.055173884564413e-05,
+      "loss": 1.0227,
+      "step": 124550
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.0550977196216126e-05,
+      "loss": 1.0023,
+      "step": 124560
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.055021554678813e-05,
+      "loss": 1.0709,
+      "step": 124570
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054945389736012e-05,
+      "loss": 1.0253,
+      "step": 124580
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.0548692247932126e-05,
+      "loss": 0.8818,
+      "step": 124590
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054793059850412e-05,
+      "loss": 0.8622,
+      "step": 124600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054716894907612e-05,
+      "loss": 0.9843,
+      "step": 124610
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054640729964812e-05,
+      "loss": 0.9403,
+      "step": 124620
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054564565022012e-05,
+      "loss": 0.9289,
+      "step": 124630
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054488400079211e-05,
+      "loss": 1.0647,
+      "step": 124640
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054412235136412e-05,
+      "loss": 1.0077,
+      "step": 124650
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.0543360701936115e-05,
+      "loss": 0.9093,
+      "step": 124660
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054259905250811e-05,
+      "loss": 0.901,
+      "step": 124670
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054183740308011e-05,
+      "loss": 1.0684,
+      "step": 124680
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.0541075753652114e-05,
+      "loss": 0.9449,
+      "step": 124690
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.054031410422411e-05,
+      "loss": 1.0107,
+      "step": 124700
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.053955245479611e-05,
+      "loss": 0.9672,
+      "step": 124710
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.053879080536811e-05,
+      "loss": 0.8838,
+      "step": 124720
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.0538029155940104e-05,
+      "loss": 1.0743,
+      "step": 124730
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.053726750651211e-05,
+      "loss": 1.0071,
+      "step": 124740
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.05365058570841e-05,
+      "loss": 0.9483,
+      "step": 124750
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.05357442076561e-05,
+      "loss": 0.9237,
+      "step": 124760
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.05349825582281e-05,
+      "loss": 0.9951,
+      "step": 124770
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.05342209088001e-05,
+      "loss": 1.0291,
+      "step": 124780
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.0533459259372095e-05,
+      "loss": 0.9411,
+      "step": 124790
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.05326976099441e-05,
+      "loss": 1.0401,
+      "step": 124800
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.053193596051609e-05,
+      "loss": 0.9632,
+      "step": 124810
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.0531174311088094e-05,
+      "loss": 1.078,
+      "step": 124820
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.053041266166009e-05,
+      "loss": 0.9735,
+      "step": 124830
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052965101223209e-05,
+      "loss": 0.9167,
+      "step": 124840
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052888936280409e-05,
+      "loss": 0.8816,
+      "step": 124850
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052812771337609e-05,
+      "loss": 0.9682,
+      "step": 124860
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052736606394809e-05,
+      "loss": 1.0415,
+      "step": 124870
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.0526604414520086e-05,
+      "loss": 0.9015,
+      "step": 124880
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.0525842765092084e-05,
+      "loss": 0.8766,
+      "step": 124890
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052508111566408e-05,
+      "loss": 0.9653,
+      "step": 124900
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.0524319466236085e-05,
+      "loss": 1.0813,
+      "step": 124910
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052355781680808e-05,
+      "loss": 1.0549,
+      "step": 124920
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.052279616738008e-05,
+      "loss": 0.8665,
+      "step": 124930
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.052203451795208e-05,
+      "loss": 1.0172,
+      "step": 124940
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.052127286852408e-05,
+      "loss": 1.0199,
+      "step": 124950
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.052051121909607e-05,
+      "loss": 0.9599,
+      "step": 124960
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051974956966808e-05,
+      "loss": 0.9944,
+      "step": 124970
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0518987920240074e-05,
+      "loss": 1.0437,
+      "step": 124980
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051822627081207e-05,
+      "loss": 0.8945,
+      "step": 124990
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051746462138407e-05,
+      "loss": 0.958,
+      "step": 125000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0516702971956073e-05,
+      "loss": 0.9169,
+      "step": 125010
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0515941322528064e-05,
+      "loss": 0.9466,
+      "step": 125020
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051517967310007e-05,
+      "loss": 0.9997,
+      "step": 125030
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0514418023672066e-05,
+      "loss": 1.1038,
+      "step": 125040
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051365637424406e-05,
+      "loss": 0.9345,
+      "step": 125050
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051289472481606e-05,
+      "loss": 0.8668,
+      "step": 125060
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0512133075388065e-05,
+      "loss": 1.0809,
+      "step": 125070
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051137142596006e-05,
+      "loss": 0.8913,
+      "step": 125080
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.051060977653206e-05,
+      "loss": 0.9653,
+      "step": 125090
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0509848127104064e-05,
+      "loss": 0.9264,
+      "step": 125100
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0509086477676055e-05,
+      "loss": 0.9717,
+      "step": 125110
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.050832482824806e-05,
+      "loss": 0.9671,
+      "step": 125120
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0507563178820057e-05,
+      "loss": 0.9645,
+      "step": 125130
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.0506801529392054e-05,
+      "loss": 0.8666,
+      "step": 125140
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.050603987996405e-05,
+      "loss": 1.0223,
+      "step": 125150
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0505278230536056e-05,
+      "loss": 0.8827,
+      "step": 125160
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0504516581108046e-05,
+      "loss": 1.0008,
+      "step": 125170
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.050375493168005e-05,
+      "loss": 0.9034,
+      "step": 125180
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.050299328225205e-05,
+      "loss": 0.9948,
+      "step": 125190
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0502231632824046e-05,
+      "loss": 0.9311,
+      "step": 125200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.050146998339604e-05,
+      "loss": 1.0149,
+      "step": 125210
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.050070833396804e-05,
+      "loss": 0.9989,
+      "step": 125220
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049994668454004e-05,
+      "loss": 0.8772,
+      "step": 125230
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049918503511204e-05,
+      "loss": 0.943,
+      "step": 125240
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049842338568404e-05,
+      "loss": 1.0259,
+      "step": 125250
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049766173625604e-05,
+      "loss": 1.0157,
+      "step": 125260
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0496900086828035e-05,
+      "loss": 1.0098,
+      "step": 125270
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049613843740003e-05,
+      "loss": 1.0056,
+      "step": 125280
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0495376787972036e-05,
+      "loss": 1.0385,
+      "step": 125290
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0494615138544034e-05,
+      "loss": 0.8904,
+      "step": 125300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049385348911603e-05,
+      "loss": 0.9595,
+      "step": 125310
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049309183968803e-05,
+      "loss": 1.0123,
+      "step": 125320
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049233019026003e-05,
+      "loss": 0.9478,
+      "step": 125330
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0491568540832024e-05,
+      "loss": 1.1369,
+      "step": 125340
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.049080689140403e-05,
+      "loss": 0.9595,
+      "step": 125350
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.0490045241976025e-05,
+      "loss": 0.9253,
+      "step": 125360
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.048928359254802e-05,
+      "loss": 0.8773,
+      "step": 125370
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.048852194312002e-05,
+      "loss": 0.9771,
+      "step": 125380
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0487760293692025e-05,
+      "loss": 0.9673,
+      "step": 125390
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0486998644264015e-05,
+      "loss": 1.0339,
+      "step": 125400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.048623699483602e-05,
+      "loss": 0.9631,
+      "step": 125410
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.048547534540802e-05,
+      "loss": 1.0583,
+      "step": 125420
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0484713695980014e-05,
+      "loss": 1.0737,
+      "step": 125430
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.048395204655201e-05,
+      "loss": 1.0509,
+      "step": 125440
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0483190397124016e-05,
+      "loss": 1.0008,
+      "step": 125450
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0482428747696014e-05,
+      "loss": 0.9177,
+      "step": 125460
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.048166709826801e-05,
+      "loss": 0.9648,
+      "step": 125470
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.048090544884001e-05,
+      "loss": 0.9333,
+      "step": 125480
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0480143799412006e-05,
+      "loss": 1.0518,
+      "step": 125490
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.047938214998401e-05,
+      "loss": 0.8975,
+      "step": 125500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.047862050055601e-05,
+      "loss": 0.9824,
+      "step": 125510
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0477858851128005e-05,
+      "loss": 1.0486,
+      "step": 125520
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.04770972017e-05,
+      "loss": 0.9556,
+      "step": 125530
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.047633555227201e-05,
+      "loss": 0.9526,
+      "step": 125540
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0475573902844e-05,
+      "loss": 0.9294,
+      "step": 125550
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0474812253416e-05,
+      "loss": 0.8947,
+      "step": 125560
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0474050603988e-05,
+      "loss": 0.9451,
+      "step": 125570
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.047328895456e-05,
+      "loss": 0.903,
+      "step": 125580
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.0472527305131994e-05,
+      "loss": 0.8658,
+      "step": 125590
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0471765655704e-05,
+      "loss": 0.9671,
+      "step": 125600
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.047100400627599e-05,
+      "loss": 0.8973,
+      "step": 125610
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0470242356847993e-05,
+      "loss": 0.9706,
+      "step": 125620
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046948070741999e-05,
+      "loss": 0.8725,
+      "step": 125630
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046871905799199e-05,
+      "loss": 0.8648,
+      "step": 125640
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0467957408563986e-05,
+      "loss": 0.9739,
+      "step": 125650
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046719575913599e-05,
+      "loss": 0.9813,
+      "step": 125660
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046643410970799e-05,
+      "loss": 0.8833,
+      "step": 125670
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0465672460279985e-05,
+      "loss": 0.906,
+      "step": 125680
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046491081085198e-05,
+      "loss": 1.0411,
+      "step": 125690
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046414916142398e-05,
+      "loss": 1.1092,
+      "step": 125700
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0463387511995984e-05,
+      "loss": 0.9101,
+      "step": 125710
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0462625862567975e-05,
+      "loss": 0.9115,
+      "step": 125720
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.046186421313998e-05,
+      "loss": 0.8801,
+      "step": 125730
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0461102563711977e-05,
+      "loss": 0.999,
+      "step": 125740
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0460340914283974e-05,
+      "loss": 0.9753,
+      "step": 125750
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.045957926485597e-05,
+      "loss": 1.0236,
+      "step": 125760
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0458817615427976e-05,
+      "loss": 1.1216,
+      "step": 125770
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0458055965999966e-05,
+      "loss": 0.9498,
+      "step": 125780
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.045729431657197e-05,
+      "loss": 1.0209,
+      "step": 125790
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.045653266714397e-05,
+      "loss": 0.9567,
+      "step": 125800
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.0455771017715966e-05,
+      "loss": 1.0756,
+      "step": 125810
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.045500936828796e-05,
+      "loss": 0.9996,
+      "step": 125820
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.045424771885997e-05,
+      "loss": 1.0152,
+      "step": 125830
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.0453486069431965e-05,
+      "loss": 0.8985,
+      "step": 125840
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.045272442000396e-05,
+      "loss": 0.9172,
+      "step": 125850
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.045196277057596e-05,
+      "loss": 0.9499,
+      "step": 125860
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.045120112114796e-05,
+      "loss": 0.9263,
+      "step": 125870
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.045043947171996e-05,
+      "loss": 0.9335,
+      "step": 125880
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044967782229196e-05,
+      "loss": 1.0065,
+      "step": 125890
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.0448916172863956e-05,
+      "loss": 1.0606,
+      "step": 125900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.0448154523435954e-05,
+      "loss": 0.9997,
+      "step": 125910
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044739287400796e-05,
+      "loss": 0.912,
+      "step": 125920
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044663122457995e-05,
+      "loss": 0.9283,
+      "step": 125930
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044586957515195e-05,
+      "loss": 0.9066,
+      "step": 125940
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044510792572395e-05,
+      "loss": 0.9578,
+      "step": 125950
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044434627629595e-05,
+      "loss": 0.8855,
+      "step": 125960
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.0443584626867945e-05,
+      "loss": 0.8668,
+      "step": 125970
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044282297743995e-05,
+      "loss": 0.9901,
+      "step": 125980
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044206132801194e-05,
+      "loss": 0.9388,
+      "step": 125990
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.0441299678583945e-05,
+      "loss": 0.8762,
+      "step": 126000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.044053802915594e-05,
+      "loss": 0.9528,
+      "step": 126010
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.043977637972794e-05,
+      "loss": 1.0587,
+      "step": 126020
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.043901473029994e-05,
+      "loss": 0.8411,
+      "step": 126030
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043825308087194e-05,
+      "loss": 0.9241,
+      "step": 126040
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043749143144394e-05,
+      "loss": 0.9081,
+      "step": 126050
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0436729782015936e-05,
+      "loss": 1.0278,
+      "step": 126060
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0435968132587934e-05,
+      "loss": 0.8776,
+      "step": 126070
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043520648315993e-05,
+      "loss": 1.0258,
+      "step": 126080
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0434444833731935e-05,
+      "loss": 0.9484,
+      "step": 126090
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043368318430393e-05,
+      "loss": 0.9745,
+      "step": 126100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043292153487593e-05,
+      "loss": 1.0095,
+      "step": 126110
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043215988544793e-05,
+      "loss": 0.9347,
+      "step": 126120
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0431398236019925e-05,
+      "loss": 1.0214,
+      "step": 126130
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.043063658659192e-05,
+      "loss": 1.0483,
+      "step": 126140
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042987493716393e-05,
+      "loss": 1.1551,
+      "step": 126150
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042911328773592e-05,
+      "loss": 1.0042,
+      "step": 126160
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042835163830792e-05,
+      "loss": 0.9225,
+      "step": 126170
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042758998887992e-05,
+      "loss": 1.0595,
+      "step": 126180
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042682833945192e-05,
+      "loss": 0.9397,
+      "step": 126190
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0426066690023914e-05,
+      "loss": 0.9775,
+      "step": 126200
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042530504059592e-05,
+      "loss": 0.8655,
+      "step": 126210
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0424543391167916e-05,
+      "loss": 0.9726,
+      "step": 126220
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.0423781741739913e-05,
+      "loss": 1.0054,
+      "step": 126230
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042302009231191e-05,
+      "loss": 0.9329,
+      "step": 126240
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.042225844288391e-05,
+      "loss": 0.9773,
+      "step": 126250
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.042149679345591e-05,
+      "loss": 0.9526,
+      "step": 126260
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.042073514402791e-05,
+      "loss": 1.1087,
+      "step": 126270
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.041997349459991e-05,
+      "loss": 0.9324,
+      "step": 126280
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.0419211845171905e-05,
+      "loss": 0.9562,
+      "step": 126290
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.041845019574391e-05,
+      "loss": 0.983,
+      "step": 126300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.04176885463159e-05,
+      "loss": 0.9923,
+      "step": 126310
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.0416926896887904e-05,
+      "loss": 0.9033,
+      "step": 126320
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.04161652474599e-05,
+      "loss": 0.8852,
+      "step": 126330
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.04154035980319e-05,
+      "loss": 0.9273,
+      "step": 126340
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.04146419486039e-05,
+      "loss": 0.9229,
+      "step": 126350
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.04138802991759e-05,
+      "loss": 0.9115,
+      "step": 126360
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.041311864974789e-05,
+      "loss": 0.8644,
+      "step": 126370
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.0412357000319896e-05,
+      "loss": 0.9318,
+      "step": 126380
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.041159535089189e-05,
+      "loss": 0.9065,
+      "step": 126390
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.041083370146389e-05,
+      "loss": 0.9554,
+      "step": 126400
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.041007205203589e-05,
+      "loss": 1.0392,
+      "step": 126410
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.040931040260789e-05,
+      "loss": 0.9434,
+      "step": 126420
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.040854875317989e-05,
+      "loss": 0.9993,
+      "step": 126430
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.040778710375189e-05,
+      "loss": 0.9237,
+      "step": 126440
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.0407025454323885e-05,
+      "loss": 0.8866,
+      "step": 126450
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.040626380489588e-05,
+      "loss": 0.8967,
+      "step": 126460
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.0405502155467887e-05,
+      "loss": 0.9727,
+      "step": 126470
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0404740506039884e-05,
+      "loss": 1.0099,
+      "step": 126480
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.040397885661188e-05,
+      "loss": 1.0364,
+      "step": 126490
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.040321720718388e-05,
+      "loss": 0.9668,
+      "step": 126500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.040245555775588e-05,
+      "loss": 0.9384,
+      "step": 126510
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0401693908327874e-05,
+      "loss": 0.9876,
+      "step": 126520
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.040093225889988e-05,
+      "loss": 0.9415,
+      "step": 126530
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0400170609471876e-05,
+      "loss": 1.0545,
+      "step": 126540
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039940896004387e-05,
+      "loss": 0.9316,
+      "step": 126550
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039864731061587e-05,
+      "loss": 1.0703,
+      "step": 126560
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0397885661187875e-05,
+      "loss": 0.9503,
+      "step": 126570
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0397124011759865e-05,
+      "loss": 1.0675,
+      "step": 126580
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039636236233187e-05,
+      "loss": 0.9256,
+      "step": 126590
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039560071290387e-05,
+      "loss": 0.9924,
+      "step": 126600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0394839063475865e-05,
+      "loss": 0.9507,
+      "step": 126610
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039407741404786e-05,
+      "loss": 1.0257,
+      "step": 126620
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039331576461986e-05,
+      "loss": 0.9047,
+      "step": 126630
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0392554115191864e-05,
+      "loss": 1.0645,
+      "step": 126640
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039179246576386e-05,
+      "loss": 0.9763,
+      "step": 126650
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.039103081633586e-05,
+      "loss": 0.9478,
+      "step": 126660
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.0390269166907856e-05,
+      "loss": 0.9004,
+      "step": 126670
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.038950751747986e-05,
+      "loss": 1.0985,
+      "step": 126680
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.038874586805185e-05,
+      "loss": 0.9451,
+      "step": 126690
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0387984218623855e-05,
+      "loss": 1.0305,
+      "step": 126700
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038722256919585e-05,
+      "loss": 0.9871,
+      "step": 126710
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038646091976785e-05,
+      "loss": 0.8412,
+      "step": 126720
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038569927033985e-05,
+      "loss": 0.8714,
+      "step": 126730
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038493762091185e-05,
+      "loss": 0.9819,
+      "step": 126740
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038417597148384e-05,
+      "loss": 1.0391,
+      "step": 126750
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038341432205585e-05,
+      "loss": 1.0887,
+      "step": 126760
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0382652672627844e-05,
+      "loss": 0.9202,
+      "step": 126770
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038189102319984e-05,
+      "loss": 0.9229,
+      "step": 126780
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.038112937377184e-05,
+      "loss": 0.9891,
+      "step": 126790
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0380367724343844e-05,
+      "loss": 0.9578,
+      "step": 126800
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.037960607491584e-05,
+      "loss": 1.0844,
+      "step": 126810
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.037884442548784e-05,
+      "loss": 1.0446,
+      "step": 126820
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0378082776059836e-05,
+      "loss": 0.8807,
+      "step": 126830
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0377321126631833e-05,
+      "loss": 0.8792,
+      "step": 126840
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.037655947720384e-05,
+      "loss": 0.8531,
+      "step": 126850
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0375797827775835e-05,
+      "loss": 0.9335,
+      "step": 126860
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.037503617834783e-05,
+      "loss": 0.9633,
+      "step": 126870
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.037427452891983e-05,
+      "loss": 0.9563,
+      "step": 126880
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0373512879491834e-05,
+      "loss": 1.0207,
+      "step": 126890
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.0372751230063825e-05,
+      "loss": 0.9212,
+      "step": 126900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.037198958063583e-05,
+      "loss": 0.8882,
+      "step": 126910
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.037122793120783e-05,
+      "loss": 1.0881,
+      "step": 126920
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.0370466281779824e-05,
+      "loss": 0.8639,
+      "step": 126930
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036970463235182e-05,
+      "loss": 0.9919,
+      "step": 126940
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.0368942982923826e-05,
+      "loss": 0.8671,
+      "step": 126950
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036818133349582e-05,
+      "loss": 0.9885,
+      "step": 126960
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036741968406782e-05,
+      "loss": 0.9862,
+      "step": 126970
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036665803463982e-05,
+      "loss": 0.9164,
+      "step": 126980
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.0365896385211816e-05,
+      "loss": 1.0565,
+      "step": 126990
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036513473578381e-05,
+      "loss": 1.0429,
+      "step": 127000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036437308635582e-05,
+      "loss": 0.9993,
+      "step": 127010
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.0363611436927815e-05,
+      "loss": 0.9324,
+      "step": 127020
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036284978749981e-05,
+      "loss": 1.0747,
+      "step": 127030
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036208813807181e-05,
+      "loss": 0.998,
+      "step": 127040
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036132648864381e-05,
+      "loss": 0.9482,
+      "step": 127050
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.036056483921581e-05,
+      "loss": 0.8923,
+      "step": 127060
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.03598031897878e-05,
+      "loss": 0.8944,
+      "step": 127070
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.0359041540359807e-05,
+      "loss": 0.9545,
+      "step": 127080
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.0358279890931804e-05,
+      "loss": 1.0091,
+      "step": 127090
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.03575182415038e-05,
+      "loss": 0.9092,
+      "step": 127100
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.03567565920758e-05,
+      "loss": 1.0361,
+      "step": 127110
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.03559949426478e-05,
+      "loss": 0.876,
+      "step": 127120
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0355233293219794e-05,
+      "loss": 1.0132,
+      "step": 127130
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.03544716437918e-05,
+      "loss": 0.945,
+      "step": 127140
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0353709994363796e-05,
+      "loss": 0.8773,
+      "step": 127150
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.035294834493579e-05,
+      "loss": 1.0444,
+      "step": 127160
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.035218669550779e-05,
+      "loss": 1.0053,
+      "step": 127170
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0351425046079795e-05,
+      "loss": 0.906,
+      "step": 127180
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0350663396651785e-05,
+      "loss": 1.0961,
+      "step": 127190
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034990174722379e-05,
+      "loss": 0.9172,
+      "step": 127200
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034914009779579e-05,
+      "loss": 0.9875,
+      "step": 127210
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0348378448367785e-05,
+      "loss": 1.0535,
+      "step": 127220
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034761679893979e-05,
+      "loss": 1.0213,
+      "step": 127230
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0346855149511786e-05,
+      "loss": 0.866,
+      "step": 127240
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0346093500083784e-05,
+      "loss": 0.9925,
+      "step": 127250
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034533185065578e-05,
+      "loss": 0.9424,
+      "step": 127260
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0344570201227786e-05,
+      "loss": 0.8568,
+      "step": 127270
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0343808551799776e-05,
+      "loss": 0.9274,
+      "step": 127280
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034304690237178e-05,
+      "loss": 1.0316,
+      "step": 127290
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034228525294378e-05,
+      "loss": 0.9145,
+      "step": 127300
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.0341523603515775e-05,
+      "loss": 0.9266,
+      "step": 127310
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034076195408777e-05,
+      "loss": 0.9743,
+      "step": 127320
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.034000030465978e-05,
+      "loss": 1.1313,
+      "step": 127330
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.033923865523177e-05,
+      "loss": 1.0755,
+      "step": 127340
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033847700580377e-05,
+      "loss": 0.9646,
+      "step": 127350
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033771535637577e-05,
+      "loss": 0.9131,
+      "step": 127360
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033695370694777e-05,
+      "loss": 0.9817,
+      "step": 127370
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0336192057519764e-05,
+      "loss": 1.0297,
+      "step": 127380
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033543040809177e-05,
+      "loss": 0.8368,
+      "step": 127390
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0334668758663766e-05,
+      "loss": 1.0026,
+      "step": 127400
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0333907109235764e-05,
+      "loss": 1.0134,
+      "step": 127410
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033314545980776e-05,
+      "loss": 0.9519,
+      "step": 127420
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033238381037976e-05,
+      "loss": 1.012,
+      "step": 127430
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033162216095176e-05,
+      "loss": 0.8864,
+      "step": 127440
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033086051152376e-05,
+      "loss": 0.9885,
+      "step": 127450
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.033009886209576e-05,
+      "loss": 0.9124,
+      "step": 127460
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0329337212667755e-05,
+      "loss": 1.0431,
+      "step": 127470
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.032857556323976e-05,
+      "loss": 0.989,
+      "step": 127480
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.032781391381175e-05,
+      "loss": 0.9496,
+      "step": 127490
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0327052264383754e-05,
+      "loss": 0.9868,
+      "step": 127500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0326290614955745e-05,
+      "loss": 0.8591,
+      "step": 127510
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.032552896552775e-05,
+      "loss": 1.009,
+      "step": 127520
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.032476731609975e-05,
+      "loss": 0.9154,
+      "step": 127530
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0324005666671744e-05,
+      "loss": 0.9586,
+      "step": 127540
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.032324401724374e-05,
+      "loss": 0.9806,
+      "step": 127550
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.0322482367815746e-05,
+      "loss": 0.9495,
+      "step": 127560
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.032172071838774e-05,
+      "loss": 1.0311,
+      "step": 127570
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.032095906895974e-05,
+      "loss": 0.9802,
+      "step": 127580
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.032019741953174e-05,
+      "loss": 1.0032,
+      "step": 127590
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.0319435770103736e-05,
+      "loss": 0.9135,
+      "step": 127600
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031867412067574e-05,
+      "loss": 0.8749,
+      "step": 127610
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031791247124774e-05,
+      "loss": 0.8759,
+      "step": 127620
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.0317150821819735e-05,
+      "loss": 1.0006,
+      "step": 127630
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031638917239173e-05,
+      "loss": 1.0194,
+      "step": 127640
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031562752296374e-05,
+      "loss": 0.8739,
+      "step": 127650
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031486587353573e-05,
+      "loss": 1.0261,
+      "step": 127660
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031410422410773e-05,
+      "loss": 0.9671,
+      "step": 127670
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031334257467973e-05,
+      "loss": 1.0922,
+      "step": 127680
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.0312580925251727e-05,
+      "loss": 0.9056,
+      "step": 127690
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.0311819275823724e-05,
+      "loss": 1.07,
+      "step": 127700
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031105762639573e-05,
+      "loss": 0.9462,
+      "step": 127710
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.031029597696772e-05,
+      "loss": 1.0162,
+      "step": 127720
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.030953432753972e-05,
+      "loss": 0.9602,
+      "step": 127730
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.030877267811172e-05,
+      "loss": 0.9597,
+      "step": 127740
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.030801102868372e-05,
+      "loss": 0.891,
+      "step": 127750
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.0307249379255716e-05,
+      "loss": 1.0118,
+      "step": 127760
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.030648772982772e-05,
+      "loss": 1.0068,
+      "step": 127770
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.030572608039971e-05,
+      "loss": 1.0774,
+      "step": 127780
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.0304964430971715e-05,
+      "loss": 1.0291,
+      "step": 127790
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.030420278154371e-05,
+      "loss": 0.9575,
+      "step": 127800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.030344113211571e-05,
+      "loss": 0.9151,
+      "step": 127810
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.0302679482687714e-05,
+      "loss": 1.003,
+      "step": 127820
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.030191783325971e-05,
+      "loss": 0.949,
+      "step": 127830
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.030115618383171e-05,
+      "loss": 0.8979,
+      "step": 127840
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.0300394534403706e-05,
+      "loss": 1.0093,
+      "step": 127850
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.029963288497571e-05,
+      "loss": 0.9892,
+      "step": 127860
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.02988712355477e-05,
+      "loss": 0.9856,
+      "step": 127870
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.0298109586119706e-05,
+      "loss": 1.0028,
+      "step": 127880
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.02973479366917e-05,
+      "loss": 1.0039,
+      "step": 127890
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.02965862872637e-05,
+      "loss": 0.8618,
+      "step": 127900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.02958246378357e-05,
+      "loss": 0.9798,
+      "step": 127910
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.02950629884077e-05,
+      "loss": 1.0045,
+      "step": 127920
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.029430133897969e-05,
+      "loss": 0.9039,
+      "step": 127930
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.02935396895517e-05,
+      "loss": 0.9261,
+      "step": 127940
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.0292778040123695e-05,
+      "loss": 0.9916,
+      "step": 127950
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.029201639069569e-05,
+      "loss": 1.0423,
+      "step": 127960
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.029125474126769e-05,
+      "loss": 0.9398,
+      "step": 127970
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.029049309183969e-05,
+      "loss": 0.9826,
+      "step": 127980
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.028973144241169e-05,
+      "loss": 0.9336,
+      "step": 127990
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.028896979298369e-05,
+      "loss": 0.9497,
+      "step": 128000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0288208143555686e-05,
+      "loss": 0.9061,
+      "step": 128010
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0287446494127684e-05,
+      "loss": 0.899,
+      "step": 128020
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028668484469969e-05,
+      "loss": 0.9577,
+      "step": 128030
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028592319527168e-05,
+      "loss": 0.9929,
+      "step": 128040
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028516154584368e-05,
+      "loss": 0.8773,
+      "step": 128050
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028439989641568e-05,
+      "loss": 0.9202,
+      "step": 128060
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028363824698768e-05,
+      "loss": 1.0732,
+      "step": 128070
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0282876597559675e-05,
+      "loss": 0.9393,
+      "step": 128080
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028211494813168e-05,
+      "loss": 1.0372,
+      "step": 128090
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.028135329870367e-05,
+      "loss": 0.9599,
+      "step": 128100
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0280591649275674e-05,
+      "loss": 0.9051,
+      "step": 128110
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027982999984767e-05,
+      "loss": 0.9935,
+      "step": 128120
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027906835041967e-05,
+      "loss": 0.8961,
+      "step": 128130
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027830670099167e-05,
+      "loss": 1.0138,
+      "step": 128140
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027754505156367e-05,
+      "loss": 1.0435,
+      "step": 128150
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027678340213566e-05,
+      "loss": 0.966,
+      "step": 128160
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0276021752707666e-05,
+      "loss": 0.9498,
+      "step": 128170
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0275260103279663e-05,
+      "loss": 1.0181,
+      "step": 128180
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027449845385166e-05,
+      "loss": 0.9779,
+      "step": 128190
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.0273736804423665e-05,
+      "loss": 0.9254,
+      "step": 128200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027297515499566e-05,
+      "loss": 0.853,
+      "step": 128210
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.027221350556766e-05,
+      "loss": 0.9869,
+      "step": 128220
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.027145185613966e-05,
+      "loss": 1.0395,
+      "step": 128230
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.027069020671166e-05,
+      "loss": 0.9683,
+      "step": 128240
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026992855728365e-05,
+      "loss": 0.9524,
+      "step": 128250
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026916690785566e-05,
+      "loss": 1.0219,
+      "step": 128260
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0268405258427654e-05,
+      "loss": 0.9663,
+      "step": 128270
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026764360899965e-05,
+      "loss": 0.822,
+      "step": 128280
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026688195957165e-05,
+      "loss": 0.9224,
+      "step": 128290
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026612031014365e-05,
+      "loss": 0.9569,
+      "step": 128300
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0265358660715644e-05,
+      "loss": 0.896,
+      "step": 128310
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026459701128765e-05,
+      "loss": 0.9456,
+      "step": 128320
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0263835361859646e-05,
+      "loss": 0.9635,
+      "step": 128330
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026307371243164e-05,
+      "loss": 0.9961,
+      "step": 128340
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026231206300364e-05,
+      "loss": 0.9388,
+      "step": 128350
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0261550413575645e-05,
+      "loss": 0.9873,
+      "step": 128360
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0260788764147636e-05,
+      "loss": 0.9711,
+      "step": 128370
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.026002711471964e-05,
+      "loss": 0.8649,
+      "step": 128380
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.025926546529164e-05,
+      "loss": 0.9816,
+      "step": 128390
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0258503815863635e-05,
+      "loss": 0.9524,
+      "step": 128400
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.025774216643564e-05,
+      "loss": 1.0049,
+      "step": 128410
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0256980517007637e-05,
+      "loss": 1.0413,
+      "step": 128420
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.0256218867579634e-05,
+      "loss": 0.9675,
+      "step": 128430
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.025545721815163e-05,
+      "loss": 0.9698,
+      "step": 128440
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.025469556872363e-05,
+      "loss": 0.9691,
+      "step": 128450
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.0253933919295626e-05,
+      "loss": 0.9605,
+      "step": 128460
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.025317226986763e-05,
+      "loss": 0.9101,
+      "step": 128470
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.025241062043962e-05,
+      "loss": 0.8979,
+      "step": 128480
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.0251648971011626e-05,
+      "loss": 0.9261,
+      "step": 128490
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.025088732158362e-05,
+      "loss": 1.0654,
+      "step": 128500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.025012567215562e-05,
+      "loss": 0.8854,
+      "step": 128510
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024936402272762e-05,
+      "loss": 0.9214,
+      "step": 128520
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024860237329962e-05,
+      "loss": 0.9409,
+      "step": 128530
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024784072387161e-05,
+      "loss": 0.9962,
+      "step": 128540
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024707907444362e-05,
+      "loss": 0.878,
+      "step": 128550
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.0246317425015615e-05,
+      "loss": 1.0613,
+      "step": 128560
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024555577558761e-05,
+      "loss": 0.9932,
+      "step": 128570
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024479412615961e-05,
+      "loss": 0.961,
+      "step": 128580
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.0244032476731614e-05,
+      "loss": 0.9546,
+      "step": 128590
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024327082730361e-05,
+      "loss": 0.9664,
+      "step": 128600
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024250917787561e-05,
+      "loss": 0.9128,
+      "step": 128610
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024174752844761e-05,
+      "loss": 1.09,
+      "step": 128620
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.0240985879019604e-05,
+      "loss": 0.9281,
+      "step": 128630
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.024022422959161e-05,
+      "loss": 1.0191,
+      "step": 128640
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.0239462580163605e-05,
+      "loss": 1.0409,
+      "step": 128650
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.02387009307356e-05,
+      "loss": 0.8822,
+      "step": 128660
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.02379392813076e-05,
+      "loss": 1.0311,
+      "step": 128670
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0237177631879605e-05,
+      "loss": 1.0006,
+      "step": 128680
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0236415982451595e-05,
+      "loss": 1.1225,
+      "step": 128690
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.02356543330236e-05,
+      "loss": 0.9982,
+      "step": 128700
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.02348926835956e-05,
+      "loss": 0.9959,
+      "step": 128710
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0234131034167594e-05,
+      "loss": 0.8635,
+      "step": 128720
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.023336938473959e-05,
+      "loss": 0.9022,
+      "step": 128730
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0232607735311596e-05,
+      "loss": 1.0552,
+      "step": 128740
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.023184608588359e-05,
+      "loss": 1.061,
+      "step": 128750
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.023108443645559e-05,
+      "loss": 1.0017,
+      "step": 128760
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.023032278702759e-05,
+      "loss": 1.0698,
+      "step": 128770
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0229561137599586e-05,
+      "loss": 0.9773,
+      "step": 128780
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022879948817159e-05,
+      "loss": 0.9897,
+      "step": 128790
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022803783874359e-05,
+      "loss": 1.0031,
+      "step": 128800
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0227276189315585e-05,
+      "loss": 1.1082,
+      "step": 128810
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022651453988758e-05,
+      "loss": 0.9862,
+      "step": 128820
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022575289045959e-05,
+      "loss": 1.0636,
+      "step": 128830
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022499124103158e-05,
+      "loss": 1.0314,
+      "step": 128840
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022422959160358e-05,
+      "loss": 0.8266,
+      "step": 128850
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022346794217558e-05,
+      "loss": 1.1222,
+      "step": 128860
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.022270629274758e-05,
+      "loss": 0.9395,
+      "step": 128870
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.0221944643319574e-05,
+      "loss": 1.0433,
+      "step": 128880
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.022118299389157e-05,
+      "loss": 0.9191,
+      "step": 128890
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.022042134446357e-05,
+      "loss": 0.9688,
+      "step": 128900
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021965969503557e-05,
+      "loss": 0.9317,
+      "step": 128910
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0218898045607564e-05,
+      "loss": 0.961,
+      "step": 128920
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021813639617957e-05,
+      "loss": 0.9424,
+      "step": 128930
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0217374746751566e-05,
+      "loss": 0.8793,
+      "step": 128940
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021661309732356e-05,
+      "loss": 0.963,
+      "step": 128950
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021585144789556e-05,
+      "loss": 0.8687,
+      "step": 128960
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0215089798467565e-05,
+      "loss": 0.9018,
+      "step": 128970
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021432814903956e-05,
+      "loss": 0.9933,
+      "step": 128980
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021356649961156e-05,
+      "loss": 0.9391,
+      "step": 128990
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0212804850183564e-05,
+      "loss": 0.9171,
+      "step": 129000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0212043200755555e-05,
+      "loss": 0.9703,
+      "step": 129010
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.021128155132756e-05,
+      "loss": 0.9217,
+      "step": 129020
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0210519901899557e-05,
+      "loss": 1.0834,
+      "step": 129030
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0209758252471554e-05,
+      "loss": 0.9527,
+      "step": 129040
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.020899660304355e-05,
+      "loss": 0.9294,
+      "step": 129050
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0208234953615556e-05,
+      "loss": 0.9555,
+      "step": 129060
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.0207473304187546e-05,
+      "loss": 0.9602,
+      "step": 129070
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.020671165475955e-05,
+      "loss": 0.9386,
+      "step": 129080
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.020595000533155e-05,
+      "loss": 0.8319,
+      "step": 129090
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.0205188355903546e-05,
+      "loss": 0.9695,
+      "step": 129100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.020442670647554e-05,
+      "loss": 1.0328,
+      "step": 129110
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.020366505704755e-05,
+      "loss": 0.9763,
+      "step": 129120
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.020290340761954e-05,
+      "loss": 0.8766,
+      "step": 129130
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.020214175819154e-05,
+      "loss": 0.8856,
+      "step": 129140
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.020138010876354e-05,
+      "loss": 0.956,
+      "step": 129150
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.020061845933554e-05,
+      "loss": 0.864,
+      "step": 129160
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.0199856809907535e-05,
+      "loss": 0.9315,
+      "step": 129170
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019909516047954e-05,
+      "loss": 0.9273,
+      "step": 129180
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.0198333511051536e-05,
+      "loss": 0.9907,
+      "step": 129190
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.0197571861623534e-05,
+      "loss": 0.9264,
+      "step": 129200
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019681021219554e-05,
+      "loss": 1.0115,
+      "step": 129210
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019604856276753e-05,
+      "loss": 0.9131,
+      "step": 129220
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019528691333953e-05,
+      "loss": 0.9288,
+      "step": 129230
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019452526391153e-05,
+      "loss": 1.0213,
+      "step": 129240
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019376361448353e-05,
+      "loss": 0.9531,
+      "step": 129250
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.0193001965055525e-05,
+      "loss": 1.0217,
+      "step": 129260
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019224031562753e-05,
+      "loss": 1.0523,
+      "step": 129270
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.019147866619952e-05,
+      "loss": 1.045,
+      "step": 129280
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.0190717016771525e-05,
+      "loss": 1.1093,
+      "step": 129290
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.018995536734352e-05,
+      "loss": 0.988,
+      "step": 129300
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.018919371791552e-05,
+      "loss": 0.9518,
+      "step": 129310
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018843206848752e-05,
+      "loss": 0.893,
+      "step": 129320
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018767041905952e-05,
+      "loss": 0.9592,
+      "step": 129330
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018690876963151e-05,
+      "loss": 1.0005,
+      "step": 129340
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.0186147120203516e-05,
+      "loss": 0.9246,
+      "step": 129350
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.0185385470775514e-05,
+      "loss": 0.8288,
+      "step": 129360
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018462382134751e-05,
+      "loss": 1.0148,
+      "step": 129370
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018386217191951e-05,
+      "loss": 1.0275,
+      "step": 129380
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.0183100522491506e-05,
+      "loss": 0.9137,
+      "step": 129390
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018233887306351e-05,
+      "loss": 0.969,
+      "step": 129400
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.018157722363551e-05,
+      "loss": 0.9706,
+      "step": 129410
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.0180815574207505e-05,
+      "loss": 0.9093,
+      "step": 129420
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.01800539247795e-05,
+      "loss": 1.009,
+      "step": 129430
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.017929227535151e-05,
+      "loss": 1.052,
+      "step": 129440
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.01785306259235e-05,
+      "loss": 0.8785,
+      "step": 129450
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.01777689764955e-05,
+      "loss": 0.9956,
+      "step": 129460
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.01770073270675e-05,
+      "loss": 0.9821,
+      "step": 129470
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.01762456776395e-05,
+      "loss": 0.9847,
+      "step": 129480
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.0175484028211494e-05,
+      "loss": 0.9531,
+      "step": 129490
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.01747223787835e-05,
+      "loss": 0.9938,
+      "step": 129500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.017396072935549e-05,
+      "loss": 0.9385,
+      "step": 129510
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.017319907992749e-05,
+      "loss": 1.0428,
+      "step": 129520
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.017243743049949e-05,
+      "loss": 0.8721,
+      "step": 129530
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.017167578107149e-05,
+      "loss": 0.9406,
+      "step": 129540
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.0170914131643486e-05,
+      "loss": 0.8897,
+      "step": 129550
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.017015248221549e-05,
+      "loss": 0.8894,
+      "step": 129560
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016939083278749e-05,
+      "loss": 1.0007,
+      "step": 129570
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.0168629183359485e-05,
+      "loss": 1.0222,
+      "step": 129580
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016786753393149e-05,
+      "loss": 1.0485,
+      "step": 129590
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016710588450348e-05,
+      "loss": 0.8892,
+      "step": 129600
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.0166344235075484e-05,
+      "loss": 0.9428,
+      "step": 129610
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016558258564748e-05,
+      "loss": 1.0026,
+      "step": 129620
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016482093621948e-05,
+      "loss": 0.9543,
+      "step": 129630
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.0164059286791477e-05,
+      "loss": 1.0226,
+      "step": 129640
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016329763736348e-05,
+      "loss": 0.9854,
+      "step": 129650
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016253598793547e-05,
+      "loss": 0.9068,
+      "step": 129660
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.0161774338507476e-05,
+      "loss": 0.9336,
+      "step": 129670
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016101268907947e-05,
+      "loss": 0.9569,
+      "step": 129680
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.016025103965147e-05,
+      "loss": 0.9018,
+      "step": 129690
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.015948939022347e-05,
+      "loss": 0.9804,
+      "step": 129700
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.015872774079547e-05,
+      "loss": 1.098,
+      "step": 129710
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.015796609136746e-05,
+      "loss": 1.0791,
+      "step": 129720
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.015720444193947e-05,
+      "loss": 0.965,
+      "step": 129730
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.0156442792511465e-05,
+      "loss": 0.9332,
+      "step": 129740
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.015568114308346e-05,
+      "loss": 0.9201,
+      "step": 129750
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.015491949365546e-05,
+      "loss": 1.0609,
+      "step": 129760
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0154157844227464e-05,
+      "loss": 0.9805,
+      "step": 129770
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.015339619479946e-05,
+      "loss": 0.9848,
+      "step": 129780
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.015263454537146e-05,
+      "loss": 1.0634,
+      "step": 129790
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.015187289594346e-05,
+      "loss": 0.9038,
+      "step": 129800
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0151111246515454e-05,
+      "loss": 0.9529,
+      "step": 129810
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.015034959708746e-05,
+      "loss": 0.9477,
+      "step": 129820
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014958794765945e-05,
+      "loss": 1.0175,
+      "step": 129830
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014882629823145e-05,
+      "loss": 0.8626,
+      "step": 129840
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014806464880345e-05,
+      "loss": 1.1103,
+      "step": 129850
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014730299937545e-05,
+      "loss": 0.8829,
+      "step": 129860
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0146541349947445e-05,
+      "loss": 0.9442,
+      "step": 129870
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014577970051945e-05,
+      "loss": 0.9141,
+      "step": 129880
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014501805109144e-05,
+      "loss": 0.9835,
+      "step": 129890
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0144256401663445e-05,
+      "loss": 0.8918,
+      "step": 129900
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014349475223544e-05,
+      "loss": 0.8577,
+      "step": 129910
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014273310280744e-05,
+      "loss": 0.9526,
+      "step": 129920
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014197145337944e-05,
+      "loss": 1.0087,
+      "step": 129930
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014120980395144e-05,
+      "loss": 0.9409,
+      "step": 129940
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.014044815452344e-05,
+      "loss": 0.9443,
+      "step": 129950
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0139686505095436e-05,
+      "loss": 0.8693,
+      "step": 129960
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0138924855667434e-05,
+      "loss": 0.9276,
+      "step": 129970
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013816320623943e-05,
+      "loss": 0.8986,
+      "step": 129980
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0137401556811435e-05,
+      "loss": 0.9101,
+      "step": 129990
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013663990738343e-05,
+      "loss": 0.8948,
+      "step": 130000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013587825795543e-05,
+      "loss": 1.0552,
+      "step": 130010
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013511660852743e-05,
+      "loss": 1.0376,
+      "step": 130020
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013435495909943e-05,
+      "loss": 0.9472,
+      "step": 130030
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013359330967142e-05,
+      "loss": 0.93,
+      "step": 130040
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013283166024343e-05,
+      "loss": 0.9967,
+      "step": 130050
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0132070010815424e-05,
+      "loss": 0.8956,
+      "step": 130060
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013130836138742e-05,
+      "loss": 0.9493,
+      "step": 130070
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.013054671195942e-05,
+      "loss": 1.03,
+      "step": 130080
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0129785062531424e-05,
+      "loss": 0.856,
+      "step": 130090
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0129023413103414e-05,
+      "loss": 0.8628,
+      "step": 130100
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.012826176367542e-05,
+      "loss": 1.0281,
+      "step": 130110
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0127500114247416e-05,
+      "loss": 0.9278,
+      "step": 130120
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.012673846481941e-05,
+      "loss": 1.0217,
+      "step": 130130
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.012597681539141e-05,
+      "loss": 0.9097,
+      "step": 130140
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0125215165963415e-05,
+      "loss": 0.9105,
+      "step": 130150
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.012445351653541e-05,
+      "loss": 0.9137,
+      "step": 130160
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.012369186710741e-05,
+      "loss": 0.9508,
+      "step": 130170
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.012293021767941e-05,
+      "loss": 0.867,
+      "step": 130180
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.0122168568251405e-05,
+      "loss": 1.0202,
+      "step": 130190
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.012140691882341e-05,
+      "loss": 0.909,
+      "step": 130200
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.012064526939541e-05,
+      "loss": 0.9562,
+      "step": 130210
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0119883619967404e-05,
+      "loss": 1.0321,
+      "step": 130220
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.01191219705394e-05,
+      "loss": 0.9988,
+      "step": 130230
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0118360321111406e-05,
+      "loss": 1.1013,
+      "step": 130240
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0117598671683397e-05,
+      "loss": 0.9459,
+      "step": 130250
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.01168370222554e-05,
+      "loss": 1.0722,
+      "step": 130260
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.01160753728274e-05,
+      "loss": 0.9571,
+      "step": 130270
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0115313723399396e-05,
+      "loss": 0.973,
+      "step": 130280
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.011455207397139e-05,
+      "loss": 0.9796,
+      "step": 130290
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.011379042454339e-05,
+      "loss": 0.958,
+      "step": 130300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.011302877511539e-05,
+      "loss": 0.9611,
+      "step": 130310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.011226712568739e-05,
+      "loss": 0.9694,
+      "step": 130320
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.011150547625939e-05,
+      "loss": 0.9673,
+      "step": 130330
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.011074382683139e-05,
+      "loss": 1.0397,
+      "step": 130340
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0109982177403385e-05,
+      "loss": 0.9519,
+      "step": 130350
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.010922052797538e-05,
+      "loss": 0.9558,
+      "step": 130360
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0108458878547386e-05,
+      "loss": 1.0116,
+      "step": 130370
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.0107697229119384e-05,
+      "loss": 1.012,
+      "step": 130380
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.010693557969138e-05,
+      "loss": 1.0153,
+      "step": 130390
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.010617393026338e-05,
+      "loss": 0.9136,
+      "step": 130400
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.010541228083538e-05,
+      "loss": 1.0094,
+      "step": 130410
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0104650631407374e-05,
+      "loss": 1.081,
+      "step": 130420
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.010388898197938e-05,
+      "loss": 1.0116,
+      "step": 130430
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0103127332551376e-05,
+      "loss": 0.9332,
+      "step": 130440
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.010236568312337e-05,
+      "loss": 0.8994,
+      "step": 130450
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.010160403369537e-05,
+      "loss": 0.8798,
+      "step": 130460
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0100842384267375e-05,
+      "loss": 0.9988,
+      "step": 130470
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0100080734839365e-05,
+      "loss": 0.9566,
+      "step": 130480
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009931908541137e-05,
+      "loss": 1.0293,
+      "step": 130490
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009855743598337e-05,
+      "loss": 0.9662,
+      "step": 130500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0097795786555365e-05,
+      "loss": 0.9415,
+      "step": 130510
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009703413712736e-05,
+      "loss": 0.8925,
+      "step": 130520
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0096272487699366e-05,
+      "loss": 0.9602,
+      "step": 130530
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0095510838271364e-05,
+      "loss": 0.8654,
+      "step": 130540
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009474918884336e-05,
+      "loss": 0.9623,
+      "step": 130550
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009398753941536e-05,
+      "loss": 1.1207,
+      "step": 130560
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0093225889987356e-05,
+      "loss": 0.9946,
+      "step": 130570
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009246424055936e-05,
+      "loss": 0.9254,
+      "step": 130580
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009170259113136e-05,
+      "loss": 0.9196,
+      "step": 130590
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0090940941703355e-05,
+      "loss": 0.93,
+      "step": 130600
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.009017929227535e-05,
+      "loss": 1.0129,
+      "step": 130610
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.008941764284736e-05,
+      "loss": 0.9502,
+      "step": 130620
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.008865599341935e-05,
+      "loss": 0.8948,
+      "step": 130630
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008789434399135e-05,
+      "loss": 1.0121,
+      "step": 130640
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008713269456335e-05,
+      "loss": 0.9849,
+      "step": 130650
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008637104513535e-05,
+      "loss": 1.0761,
+      "step": 130660
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0085609395707344e-05,
+      "loss": 0.9048,
+      "step": 130670
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008484774627935e-05,
+      "loss": 0.9328,
+      "step": 130680
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008408609685134e-05,
+      "loss": 0.9395,
+      "step": 130690
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0083324447423344e-05,
+      "loss": 0.9814,
+      "step": 130700
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008256279799534e-05,
+      "loss": 1.0282,
+      "step": 130710
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.008180114856734e-05,
+      "loss": 0.8242,
+      "step": 130720
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0081039499139336e-05,
+      "loss": 0.8896,
+      "step": 130730
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0080277849711333e-05,
+      "loss": 1.0144,
+      "step": 130740
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.007951620028334e-05,
+      "loss": 0.9289,
+      "step": 130750
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0078754550855335e-05,
+      "loss": 1.0387,
+      "step": 130760
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.007799290142733e-05,
+      "loss": 1.0289,
+      "step": 130770
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.007723125199933e-05,
+      "loss": 0.9565,
+      "step": 130780
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0076469602571334e-05,
+      "loss": 0.9937,
+      "step": 130790
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0075707953143325e-05,
+      "loss": 0.8766,
+      "step": 130800
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.007494630371533e-05,
+      "loss": 0.9852,
+      "step": 130810
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.007418465428733e-05,
+      "loss": 1.0059,
+      "step": 130820
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0073423004859324e-05,
+      "loss": 1.0186,
+      "step": 130830
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.007266135543132e-05,
+      "loss": 0.9624,
+      "step": 130840
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0071899706003326e-05,
+      "loss": 1.0114,
+      "step": 130850
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0071138056575317e-05,
+      "loss": 0.975,
+      "step": 130860
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.007037640714732e-05,
+      "loss": 1.0131,
+      "step": 130870
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006961475771932e-05,
+      "loss": 1.0097,
+      "step": 130880
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0068853108291316e-05,
+      "loss": 0.9907,
+      "step": 130890
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006809145886331e-05,
+      "loss": 0.95,
+      "step": 130900
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006732980943532e-05,
+      "loss": 0.9443,
+      "step": 130910
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0066568160007315e-05,
+      "loss": 0.9583,
+      "step": 130920
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006580651057931e-05,
+      "loss": 0.946,
+      "step": 130930
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006504486115131e-05,
+      "loss": 0.8972,
+      "step": 130940
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006428321172331e-05,
+      "loss": 0.961,
+      "step": 130950
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006352156229531e-05,
+      "loss": 1.0245,
+      "step": 130960
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006275991286731e-05,
+      "loss": 0.9483,
+      "step": 130970
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0061998263439306e-05,
+      "loss": 1.0891,
+      "step": 130980
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0061236614011304e-05,
+      "loss": 1.0202,
+      "step": 130990
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.006047496458331e-05,
+      "loss": 0.903,
+      "step": 131000
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.00597133151553e-05,
+      "loss": 0.9506,
+      "step": 131010
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.00589516657273e-05,
+      "loss": 1.0206,
+      "step": 131020
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.00581900162993e-05,
+      "loss": 0.9236,
+      "step": 131030
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.00574283668713e-05,
+      "loss": 0.913,
+      "step": 131040
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.0056666717443296e-05,
+      "loss": 1.0072,
+      "step": 131050
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.00559050680153e-05,
+      "loss": 0.8953,
+      "step": 131060
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.005514341858729e-05,
+      "loss": 1.0341,
+      "step": 131070
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0054381769159295e-05,
+      "loss": 0.891,
+      "step": 131080
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.005362011973129e-05,
+      "loss": 1.0107,
+      "step": 131090
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.005285847030329e-05,
+      "loss": 0.8774,
+      "step": 131100
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.005209682087529e-05,
+      "loss": 0.834,
+      "step": 131110
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.005133517144729e-05,
+      "loss": 0.9847,
+      "step": 131120
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.005057352201929e-05,
+      "loss": 0.9085,
+      "step": 131130
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0049811872591286e-05,
+      "loss": 0.9632,
+      "step": 131140
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0049050223163284e-05,
+      "loss": 0.9866,
+      "step": 131150
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004828857373528e-05,
+      "loss": 0.9254,
+      "step": 131160
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0047526924307285e-05,
+      "loss": 0.9651,
+      "step": 131170
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004676527487928e-05,
+      "loss": 0.9924,
+      "step": 131180
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004600362545128e-05,
+      "loss": 0.9035,
+      "step": 131190
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004524197602328e-05,
+      "loss": 1.062,
+      "step": 131200
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0044480326595275e-05,
+      "loss": 0.9978,
+      "step": 131210
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004371867716727e-05,
+      "loss": 0.9329,
+      "step": 131220
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004295702773928e-05,
+      "loss": 0.9199,
+      "step": 131230
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004219537831127e-05,
+      "loss": 0.9913,
+      "step": 131240
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004143372888327e-05,
+      "loss": 0.9915,
+      "step": 131250
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.004067207945527e-05,
+      "loss": 0.9126,
+      "step": 131260
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.003991043002727e-05,
+      "loss": 0.8348,
+      "step": 131270
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.0039148780599264e-05,
+      "loss": 0.972,
+      "step": 131280
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003838713117127e-05,
+      "loss": 0.9979,
+      "step": 131290
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.0037625481743266e-05,
+      "loss": 1.0585,
+      "step": 131300
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.0036863832315264e-05,
+      "loss": 0.8875,
+      "step": 131310
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003610218288726e-05,
+      "loss": 0.9318,
+      "step": 131320
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003534053345926e-05,
+      "loss": 0.9188,
+      "step": 131330
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003457888403126e-05,
+      "loss": 1.0127,
+      "step": 131340
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003381723460326e-05,
+      "loss": 0.954,
+      "step": 131350
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003305558517526e-05,
+      "loss": 0.8781,
+      "step": 131360
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.0032293935747255e-05,
+      "loss": 0.945,
+      "step": 131370
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003153228631926e-05,
+      "loss": 0.9991,
+      "step": 131380
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.003077063689125e-05,
+      "loss": 0.9474,
+      "step": 131390
+    },
+    {
+      "epoch": 6.0,
+      "eval_cer": 0.6240789139028141,
+      "eval_em": 0.007379375591296121,
+      "eval_f1": 0.007379375591296121,
+      "eval_loss": 0.9074851870536804,
+      "eval_runtime": 1010.3492,
+      "eval_samples_per_second": 10.462,
+      "eval_steps_per_second": 1.308,
+      "eval_wer": 0.9926206244087039,
+      "step": 131398
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.0030008987463254e-05,
+      "loss": 0.9348,
+      "step": 131400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002924733803525e-05,
+      "loss": 0.8592,
+      "step": 131410
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002848568860725e-05,
+      "loss": 0.9952,
+      "step": 131420
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002772403917925e-05,
+      "loss": 0.8753,
+      "step": 131430
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002696238975125e-05,
+      "loss": 0.9157,
+      "step": 131440
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002620074032324e-05,
+      "loss": 0.9012,
+      "step": 131450
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.0025439090895246e-05,
+      "loss": 0.9304,
+      "step": 131460
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002467744146724e-05,
+      "loss": 0.9139,
+      "step": 131470
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002391579203924e-05,
+      "loss": 0.8945,
+      "step": 131480
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002315414261124e-05,
+      "loss": 1.0339,
+      "step": 131490
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.002239249318324e-05,
+      "loss": 0.9606,
+      "step": 131500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.002163084375524e-05,
+      "loss": 0.908,
+      "step": 131510
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.002086919432724e-05,
+      "loss": 1.071,
+      "step": 131520
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0020107544899235e-05,
+      "loss": 0.9145,
+      "step": 131530
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001934589547123e-05,
+      "loss": 1.0437,
+      "step": 131540
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001858424604324e-05,
+      "loss": 0.9039,
+      "step": 131550
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0017822596615234e-05,
+      "loss": 0.9254,
+      "step": 131560
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001706094718723e-05,
+      "loss": 0.88,
+      "step": 131570
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001629929775923e-05,
+      "loss": 0.8171,
+      "step": 131580
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001553764833123e-05,
+      "loss": 0.8546,
+      "step": 131590
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0014775998903224e-05,
+      "loss": 0.9745,
+      "step": 131600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001401434947523e-05,
+      "loss": 0.9119,
+      "step": 131610
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0013252700047226e-05,
+      "loss": 0.9024,
+      "step": 131620
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001249105061922e-05,
+      "loss": 0.9549,
+      "step": 131630
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.001172940119122e-05,
+      "loss": 0.9555,
+      "step": 131640
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0010967751763225e-05,
+      "loss": 0.968,
+      "step": 131650
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0010206102335216e-05,
+      "loss": 0.8415,
+      "step": 131660
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.000944445290722e-05,
+      "loss": 0.9339,
+      "step": 131670
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.000868280347921e-05,
+      "loss": 1.049,
+      "step": 131680
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0007921154051215e-05,
+      "loss": 0.9816,
+      "step": 131690
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.000715950462321e-05,
+      "loss": 0.9889,
+      "step": 131700
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.000639785519521e-05,
+      "loss": 0.9779,
+      "step": 131710
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0005636205767214e-05,
+      "loss": 1.0372,
+      "step": 131720
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.000487455633921e-05,
+      "loss": 0.9413,
+      "step": 131730
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.000411290691121e-05,
+      "loss": 0.9416,
+      "step": 131740
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.0003351257483206e-05,
+      "loss": 0.9336,
+      "step": 131750
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.000258960805521e-05,
+      "loss": 1.0276,
+      "step": 131760
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.00018279586272e-05,
+      "loss": 0.8124,
+      "step": 131770
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.0001066309199205e-05,
+      "loss": 1.0081,
+      "step": 131780
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.00003046597712e-05,
+      "loss": 0.8857,
+      "step": 131790
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.99995430103432e-05,
+      "loss": 0.8715,
+      "step": 131800
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.99987813609152e-05,
+      "loss": 0.9651,
+      "step": 131810
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.99980197114872e-05,
+      "loss": 0.9039,
+      "step": 131820
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.999725806205919e-05,
+      "loss": 0.9295,
+      "step": 131830
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.99964964126312e-05,
+      "loss": 0.9349,
+      "step": 131840
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.9995734763203195e-05,
+      "loss": 0.9249,
+      "step": 131850
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.999497311377519e-05,
+      "loss": 0.9436,
+      "step": 131860
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.999421146434719e-05,
+      "loss": 0.8961,
+      "step": 131870
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.9993449814919194e-05,
+      "loss": 0.9066,
+      "step": 131880
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.999268816549119e-05,
+      "loss": 0.8729,
+      "step": 131890
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.999192651606319e-05,
+      "loss": 0.897,
+      "step": 131900
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.9991164866635186e-05,
+      "loss": 0.764,
+      "step": 131910
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.9990403217207184e-05,
+      "loss": 0.8275,
+      "step": 131920
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.998964156777919e-05,
+      "loss": 0.9516,
+      "step": 131930
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.9988879918351185e-05,
+      "loss": 0.9556,
+      "step": 131940
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998811826892318e-05,
+      "loss": 0.9431,
+      "step": 131950
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998735661949518e-05,
+      "loss": 0.8921,
+      "step": 131960
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9986594970067184e-05,
+      "loss": 0.9982,
+      "step": 131970
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9985833320639175e-05,
+      "loss": 0.8758,
+      "step": 131980
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998507167121118e-05,
+      "loss": 0.8675,
+      "step": 131990
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998431002178318e-05,
+      "loss": 1.0587,
+      "step": 132000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9983548372355174e-05,
+      "loss": 0.8329,
+      "step": 132010
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998278672292717e-05,
+      "loss": 1.0407,
+      "step": 132020
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9982025073499176e-05,
+      "loss": 0.8882,
+      "step": 132030
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998126342407117e-05,
+      "loss": 0.8924,
+      "step": 132040
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.998050177464317e-05,
+      "loss": 0.9958,
+      "step": 132050
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997974012521517e-05,
+      "loss": 1.0196,
+      "step": 132060
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9978978475787166e-05,
+      "loss": 0.8721,
+      "step": 132070
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997821682635916e-05,
+      "loss": 0.9226,
+      "step": 132080
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997745517693117e-05,
+      "loss": 1.1257,
+      "step": 132090
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9976693527503165e-05,
+      "loss": 0.8827,
+      "step": 132100
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997593187807516e-05,
+      "loss": 0.9067,
+      "step": 132110
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997517022864716e-05,
+      "loss": 0.8494,
+      "step": 132120
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997440857921916e-05,
+      "loss": 0.9937,
+      "step": 132130
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997364692979116e-05,
+      "loss": 0.9094,
+      "step": 132140
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997288528036315e-05,
+      "loss": 1.009,
+      "step": 132150
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.997212363093516e-05,
+      "loss": 1.028,
+      "step": 132160
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9971361981507154e-05,
+      "loss": 0.9417,
+      "step": 132170
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.997060033207915e-05,
+      "loss": 0.9693,
+      "step": 132180
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996983868265115e-05,
+      "loss": 0.9231,
+      "step": 132190
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996907703322315e-05,
+      "loss": 0.8815,
+      "step": 132200
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9968315383795144e-05,
+      "loss": 0.8541,
+      "step": 132210
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996755373436715e-05,
+      "loss": 0.8821,
+      "step": 132220
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9966792084939146e-05,
+      "loss": 0.917,
+      "step": 132230
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996603043551114e-05,
+      "loss": 0.9173,
+      "step": 132240
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996526878608314e-05,
+      "loss": 0.8656,
+      "step": 132250
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9964507136655145e-05,
+      "loss": 0.9737,
+      "step": 132260
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9963745487227136e-05,
+      "loss": 1.0355,
+      "step": 132270
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996298383779914e-05,
+      "loss": 0.9773,
+      "step": 132280
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996222218837114e-05,
+      "loss": 1.0515,
+      "step": 132290
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9961460538943135e-05,
+      "loss": 0.9795,
+      "step": 132300
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.996069888951514e-05,
+      "loss": 0.9097,
+      "step": 132310
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9959937240087136e-05,
+      "loss": 1.0426,
+      "step": 132320
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9959175590659134e-05,
+      "loss": 1.111,
+      "step": 132330
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.995841394123113e-05,
+      "loss": 1.0699,
+      "step": 132340
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9957652291803136e-05,
+      "loss": 1.0045,
+      "step": 132350
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9956890642375126e-05,
+      "loss": 0.8791,
+      "step": 132360
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.995612899294713e-05,
+      "loss": 0.9744,
+      "step": 132370
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.995536734351913e-05,
+      "loss": 0.9331,
+      "step": 132380
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.9954605694091125e-05,
+      "loss": 0.9179,
+      "step": 132390
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.995384404466312e-05,
+      "loss": 0.9452,
+      "step": 132400
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.995308239523513e-05,
+      "loss": 0.8634,
+      "step": 132410
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.995232074580712e-05,
+      "loss": 0.8965,
+      "step": 132420
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.995155909637912e-05,
+      "loss": 1.0045,
+      "step": 132430
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.995079744695112e-05,
+      "loss": 0.9105,
+      "step": 132440
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.995003579752312e-05,
+      "loss": 1.0225,
+      "step": 132450
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.9949274148095115e-05,
+      "loss": 1.0125,
+      "step": 132460
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994851249866712e-05,
+      "loss": 0.8972,
+      "step": 132470
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994775084923911e-05,
+      "loss": 0.972,
+      "step": 132480
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.9946989199811114e-05,
+      "loss": 0.951,
+      "step": 132490
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994622755038311e-05,
+      "loss": 0.8907,
+      "step": 132500
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994546590095511e-05,
+      "loss": 0.9581,
+      "step": 132510
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994470425152711e-05,
+      "loss": 0.8884,
+      "step": 132520
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994394260209911e-05,
+      "loss": 1.0265,
+      "step": 132530
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994318095267111e-05,
+      "loss": 0.9905,
+      "step": 132540
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.9942419303243105e-05,
+      "loss": 0.9576,
+      "step": 132550
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.994165765381511e-05,
+      "loss": 0.8447,
+      "step": 132560
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.99408960043871e-05,
+      "loss": 1.0068,
+      "step": 132570
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.9940134354959104e-05,
+      "loss": 0.8391,
+      "step": 132580
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.9939372705531095e-05,
+      "loss": 0.9073,
+      "step": 132590
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.99386110561031e-05,
+      "loss": 0.8738,
+      "step": 132600
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.99378494066751e-05,
+      "loss": 0.9801,
+      "step": 132610
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.9937087757247094e-05,
+      "loss": 0.9722,
+      "step": 132620
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.993632610781909e-05,
+      "loss": 0.9129,
+      "step": 132630
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.9935564458391096e-05,
+      "loss": 0.9761,
+      "step": 132640
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.993480280896309e-05,
+      "loss": 1.0224,
+      "step": 132650
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.993404115953509e-05,
+      "loss": 1.0176,
+      "step": 132660
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.993327951010709e-05,
+      "loss": 0.8754,
+      "step": 132670
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.9932517860679086e-05,
+      "loss": 0.8978,
+      "step": 132680
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.993175621125109e-05,
+      "loss": 1.0176,
+      "step": 132690
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.993099456182309e-05,
+      "loss": 0.968,
+      "step": 132700
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.9930232912395085e-05,
+      "loss": 0.9891,
+      "step": 132710
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992947126296708e-05,
+      "loss": 0.9761,
+      "step": 132720
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992870961353909e-05,
+      "loss": 1.0251,
+      "step": 132730
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992794796411108e-05,
+      "loss": 1.0228,
+      "step": 132740
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992718631468308e-05,
+      "loss": 0.8865,
+      "step": 132750
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992642466525508e-05,
+      "loss": 0.8738,
+      "step": 132760
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992566301582708e-05,
+      "loss": 0.8764,
+      "step": 132770
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.9924901366399074e-05,
+      "loss": 1.0952,
+      "step": 132780
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992413971697108e-05,
+      "loss": 0.955,
+      "step": 132790
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992337806754307e-05,
+      "loss": 0.9545,
+      "step": 132800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992261641811507e-05,
+      "loss": 0.8633,
+      "step": 132810
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.992185476868707e-05,
+      "loss": 1.1055,
+      "step": 132820
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.992109311925907e-05,
+      "loss": 0.973,
+      "step": 132830
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9920331469831066e-05,
+      "loss": 0.8864,
+      "step": 132840
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991956982040307e-05,
+      "loss": 0.8371,
+      "step": 132850
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991880817097506e-05,
+      "loss": 0.9455,
+      "step": 132860
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9918046521547065e-05,
+      "loss": 0.9232,
+      "step": 132870
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991728487211906e-05,
+      "loss": 1.0404,
+      "step": 132880
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991652322269106e-05,
+      "loss": 0.986,
+      "step": 132890
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9915761573263064e-05,
+      "loss": 0.9119,
+      "step": 132900
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991499992383506e-05,
+      "loss": 0.8867,
+      "step": 132910
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991423827440706e-05,
+      "loss": 0.9573,
+      "step": 132920
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9913476624979056e-05,
+      "loss": 0.9159,
+      "step": 132930
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991271497555106e-05,
+      "loss": 0.999,
+      "step": 132940
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991195332612305e-05,
+      "loss": 0.871,
+      "step": 132950
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9911191676695056e-05,
+      "loss": 0.8911,
+      "step": 132960
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.991043002726705e-05,
+      "loss": 1.0361,
+      "step": 132970
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.990966837783905e-05,
+      "loss": 1.0206,
+      "step": 132980
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.990890672841105e-05,
+      "loss": 0.9613,
+      "step": 132990
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.990814507898305e-05,
+      "loss": 0.905,
+      "step": 133000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.990738342955504e-05,
+      "loss": 0.9161,
+      "step": 133010
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.990662178012705e-05,
+      "loss": 0.9955,
+      "step": 133020
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9905860130699045e-05,
+      "loss": 0.9481,
+      "step": 133030
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.990509848127104e-05,
+      "loss": 0.9765,
+      "step": 133040
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.990433683184304e-05,
+      "loss": 0.9925,
+      "step": 133050
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.990357518241504e-05,
+      "loss": 0.9737,
+      "step": 133060
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9902813532987035e-05,
+      "loss": 0.8866,
+      "step": 133070
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.990205188355904e-05,
+      "loss": 0.9458,
+      "step": 133080
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9901290234131036e-05,
+      "loss": 0.9093,
+      "step": 133090
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9900528584703034e-05,
+      "loss": 0.9356,
+      "step": 133100
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989976693527504e-05,
+      "loss": 0.9395,
+      "step": 133110
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989900528584703e-05,
+      "loss": 0.9598,
+      "step": 133120
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989824363641903e-05,
+      "loss": 0.8956,
+      "step": 133130
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989748198699103e-05,
+      "loss": 0.9181,
+      "step": 133140
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989672033756303e-05,
+      "loss": 0.9774,
+      "step": 133150
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9895958688135025e-05,
+      "loss": 0.9823,
+      "step": 133160
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989519703870703e-05,
+      "loss": 0.9604,
+      "step": 133170
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989443538927902e-05,
+      "loss": 0.964,
+      "step": 133180
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9893673739851024e-05,
+      "loss": 0.8606,
+      "step": 133190
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989291209042302e-05,
+      "loss": 0.8469,
+      "step": 133200
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989215044099502e-05,
+      "loss": 0.9999,
+      "step": 133210
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989138879156702e-05,
+      "loss": 0.9486,
+      "step": 133220
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.989062714213902e-05,
+      "loss": 0.8901,
+      "step": 133230
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.988986549271101e-05,
+      "loss": 0.8989,
+      "step": 133240
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9889103843283016e-05,
+      "loss": 0.9194,
+      "step": 133250
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9888342193855014e-05,
+      "loss": 0.8966,
+      "step": 133260
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988758054442701e-05,
+      "loss": 0.9895,
+      "step": 133270
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988681889499901e-05,
+      "loss": 0.8347,
+      "step": 133280
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988605724557101e-05,
+      "loss": 0.8997,
+      "step": 133290
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988529559614301e-05,
+      "loss": 0.9619,
+      "step": 133300
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988453394671501e-05,
+      "loss": 0.9381,
+      "step": 133310
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988377229728701e-05,
+      "loss": 0.9842,
+      "step": 133320
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9883010647859e-05,
+      "loss": 1.027,
+      "step": 133330
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.988224899843101e-05,
+      "loss": 0.8477,
+      "step": 133340
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9881487349003004e-05,
+      "loss": 0.9378,
+      "step": 133350
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9880725699575e-05,
+      "loss": 0.9439,
+      "step": 133360
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9879964050147e-05,
+      "loss": 0.9059,
+      "step": 133370
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9879202400719003e-05,
+      "loss": 0.8116,
+      "step": 133380
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9878440751290994e-05,
+      "loss": 0.9534,
+      "step": 133390
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9877679101863e-05,
+      "loss": 0.8203,
+      "step": 133400
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9876917452434996e-05,
+      "loss": 1.0061,
+      "step": 133410
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.987615580300699e-05,
+      "loss": 1.0369,
+      "step": 133420
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.987539415357899e-05,
+      "loss": 0.9045,
+      "step": 133430
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9874632504150995e-05,
+      "loss": 0.9495,
+      "step": 133440
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9873870854722986e-05,
+      "loss": 1.0227,
+      "step": 133450
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.987310920529499e-05,
+      "loss": 0.8813,
+      "step": 133460
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.987234755586699e-05,
+      "loss": 0.9937,
+      "step": 133470
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.9871585906438985e-05,
+      "loss": 0.8284,
+      "step": 133480
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.987082425701099e-05,
+      "loss": 0.9971,
+      "step": 133490
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.987006260758298e-05,
+      "loss": 1.0187,
+      "step": 133500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.9869300958154984e-05,
+      "loss": 0.9538,
+      "step": 133510
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986853930872698e-05,
+      "loss": 0.9368,
+      "step": 133520
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986777765929898e-05,
+      "loss": 0.917,
+      "step": 133530
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.9867016009870976e-05,
+      "loss": 1.0257,
+      "step": 133540
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986625436044298e-05,
+      "loss": 0.8845,
+      "step": 133550
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986549271101497e-05,
+      "loss": 0.9484,
+      "step": 133560
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.9864731061586976e-05,
+      "loss": 0.9203,
+      "step": 133570
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986396941215897e-05,
+      "loss": 1.003,
+      "step": 133580
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986320776273097e-05,
+      "loss": 0.9194,
+      "step": 133590
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986244611330297e-05,
+      "loss": 1.0536,
+      "step": 133600
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986168446387497e-05,
+      "loss": 1.018,
+      "step": 133610
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986092281444696e-05,
+      "loss": 0.8578,
+      "step": 133620
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.986016116501897e-05,
+      "loss": 0.9265,
+      "step": 133630
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.9859399515590965e-05,
+      "loss": 0.9134,
+      "step": 133640
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.985863786616296e-05,
+      "loss": 0.9945,
+      "step": 133650
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.985787621673496e-05,
+      "loss": 0.8403,
+      "step": 133660
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.9857114567306964e-05,
+      "loss": 0.9868,
+      "step": 133670
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.985635291787896e-05,
+      "loss": 1.045,
+      "step": 133680
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.985559126845096e-05,
+      "loss": 0.9306,
+      "step": 133690
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.985482961902296e-05,
+      "loss": 0.9288,
+      "step": 133700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9854067969594954e-05,
+      "loss": 0.8986,
+      "step": 133710
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.985330632016696e-05,
+      "loss": 0.8611,
+      "step": 133720
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9852544670738955e-05,
+      "loss": 0.9243,
+      "step": 133730
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.985178302131095e-05,
+      "loss": 0.8468,
+      "step": 133740
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.985102137188295e-05,
+      "loss": 0.9651,
+      "step": 133750
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9850259722454955e-05,
+      "loss": 0.8657,
+      "step": 133760
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9849498073026945e-05,
+      "loss": 0.9371,
+      "step": 133770
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984873642359895e-05,
+      "loss": 0.9054,
+      "step": 133780
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984797477417095e-05,
+      "loss": 0.9695,
+      "step": 133790
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9847213124742944e-05,
+      "loss": 0.9925,
+      "step": 133800
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984645147531494e-05,
+      "loss": 0.9449,
+      "step": 133810
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9845689825886946e-05,
+      "loss": 0.993,
+      "step": 133820
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984492817645894e-05,
+      "loss": 0.8277,
+      "step": 133830
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984416652703094e-05,
+      "loss": 0.8578,
+      "step": 133840
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984340487760294e-05,
+      "loss": 0.9936,
+      "step": 133850
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9842643228174936e-05,
+      "loss": 0.9476,
+      "step": 133860
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9841881578746934e-05,
+      "loss": 0.9084,
+      "step": 133870
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.984111992931894e-05,
+      "loss": 0.9032,
+      "step": 133880
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.9840358279890935e-05,
+      "loss": 1.0209,
+      "step": 133890
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.983959663046293e-05,
+      "loss": 0.8187,
+      "step": 133900
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.983883498103494e-05,
+      "loss": 0.8957,
+      "step": 133910
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983807333160693e-05,
+      "loss": 0.8622,
+      "step": 133920
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983731168217893e-05,
+      "loss": 0.9302,
+      "step": 133930
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983655003275093e-05,
+      "loss": 0.8427,
+      "step": 133940
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983578838332293e-05,
+      "loss": 0.9782,
+      "step": 133950
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9835026733894924e-05,
+      "loss": 0.859,
+      "step": 133960
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983426508446692e-05,
+      "loss": 0.9426,
+      "step": 133970
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983350343503892e-05,
+      "loss": 0.8408,
+      "step": 133980
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9832741785610923e-05,
+      "loss": 0.9025,
+      "step": 133990
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9831980136182914e-05,
+      "loss": 0.9761,
+      "step": 134000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.983121848675492e-05,
+      "loss": 0.9267,
+      "step": 134010
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9830456837326916e-05,
+      "loss": 1.0129,
+      "step": 134020
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.982969518789891e-05,
+      "loss": 1.1043,
+      "step": 134030
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.982893353847091e-05,
+      "loss": 0.989,
+      "step": 134040
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9828171889042915e-05,
+      "loss": 1.1093,
+      "step": 134050
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.982741023961491e-05,
+      "loss": 1.0129,
+      "step": 134060
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.982664859018691e-05,
+      "loss": 0.9939,
+      "step": 134070
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.982588694075891e-05,
+      "loss": 0.908,
+      "step": 134080
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9825125291330905e-05,
+      "loss": 0.9895,
+      "step": 134090
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.982436364190291e-05,
+      "loss": 0.9443,
+      "step": 134100
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9823601992474907e-05,
+      "loss": 0.9169,
+      "step": 134110
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.9822840343046904e-05,
+      "loss": 0.8003,
+      "step": 134120
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.98220786936189e-05,
+      "loss": 0.9841,
+      "step": 134130
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9821317044190906e-05,
+      "loss": 1.0093,
+      "step": 134140
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9820555394762896e-05,
+      "loss": 0.9599,
+      "step": 134150
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.98197937453349e-05,
+      "loss": 0.8548,
+      "step": 134160
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.98190320959069e-05,
+      "loss": 1.0035,
+      "step": 134170
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9818270446478896e-05,
+      "loss": 0.9491,
+      "step": 134180
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.981750879705089e-05,
+      "loss": 0.9427,
+      "step": 134190
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.98167471476229e-05,
+      "loss": 0.8509,
+      "step": 134200
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.981598549819489e-05,
+      "loss": 0.7994,
+      "step": 134210
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.981522384876689e-05,
+      "loss": 0.9303,
+      "step": 134220
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.981446219933889e-05,
+      "loss": 0.935,
+      "step": 134230
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.981370054991089e-05,
+      "loss": 0.972,
+      "step": 134240
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9812938900482885e-05,
+      "loss": 0.8165,
+      "step": 134250
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.981217725105489e-05,
+      "loss": 0.906,
+      "step": 134260
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9811415601626886e-05,
+      "loss": 0.9681,
+      "step": 134270
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9810653952198884e-05,
+      "loss": 0.8315,
+      "step": 134280
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.980989230277089e-05,
+      "loss": 0.9829,
+      "step": 134290
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.980913065334288e-05,
+      "loss": 0.9106,
+      "step": 134300
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.980836900391488e-05,
+      "loss": 0.9954,
+      "step": 134310
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.980760735448688e-05,
+      "loss": 0.9464,
+      "step": 134320
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.980684570505888e-05,
+      "loss": 0.8846,
+      "step": 134330
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.9806084055630875e-05,
+      "loss": 0.9938,
+      "step": 134340
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.980532240620288e-05,
+      "loss": 1.0596,
+      "step": 134350
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.980456075677487e-05,
+      "loss": 0.9832,
+      "step": 134360
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.9803799107346875e-05,
+      "loss": 0.9337,
+      "step": 134370
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.980303745791887e-05,
+      "loss": 0.9255,
+      "step": 134380
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.980227580849087e-05,
+      "loss": 0.9356,
+      "step": 134390
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.980151415906287e-05,
+      "loss": 1.0154,
+      "step": 134400
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.980075250963487e-05,
+      "loss": 0.8868,
+      "step": 134410
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979999086020686e-05,
+      "loss": 0.9424,
+      "step": 134420
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.9799229210778866e-05,
+      "loss": 1.0448,
+      "step": 134430
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.9798467561350864e-05,
+      "loss": 0.9587,
+      "step": 134440
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979770591192286e-05,
+      "loss": 0.9046,
+      "step": 134450
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979694426249486e-05,
+      "loss": 0.9058,
+      "step": 134460
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.9796182613066856e-05,
+      "loss": 0.9568,
+      "step": 134470
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979542096363886e-05,
+      "loss": 1.0327,
+      "step": 134480
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979465931421086e-05,
+      "loss": 1.1241,
+      "step": 134490
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.9793897664782855e-05,
+      "loss": 0.9741,
+      "step": 134500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979313601535485e-05,
+      "loss": 0.9549,
+      "step": 134510
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979237436592686e-05,
+      "loss": 0.9243,
+      "step": 134520
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979161271649885e-05,
+      "loss": 0.8683,
+      "step": 134530
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979085106707085e-05,
+      "loss": 1.051,
+      "step": 134540
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.979008941764285e-05,
+      "loss": 0.9535,
+      "step": 134550
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.978932776821485e-05,
+      "loss": 1.0279,
+      "step": 134560
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.9788566118786844e-05,
+      "loss": 0.9668,
+      "step": 134570
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978780446935885e-05,
+      "loss": 0.993,
+      "step": 134580
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978704281993084e-05,
+      "loss": 0.9759,
+      "step": 134590
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.9786281170502843e-05,
+      "loss": 0.9662,
+      "step": 134600
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978551952107484e-05,
+      "loss": 0.897,
+      "step": 134610
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978475787164684e-05,
+      "loss": 0.882,
+      "step": 134620
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.9783996222218836e-05,
+      "loss": 0.941,
+      "step": 134630
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978323457279084e-05,
+      "loss": 0.9217,
+      "step": 134640
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978247292336284e-05,
+      "loss": 0.8392,
+      "step": 134650
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.9781711273934835e-05,
+      "loss": 0.9088,
+      "step": 134660
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978094962450683e-05,
+      "loss": 0.9718,
+      "step": 134670
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.978018797507883e-05,
+      "loss": 0.9291,
+      "step": 134680
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.9779426325650834e-05,
+      "loss": 0.9487,
+      "step": 134690
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977866467622283e-05,
+      "loss": 1.0165,
+      "step": 134700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977790302679483e-05,
+      "loss": 0.8229,
+      "step": 134710
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.9777141377366827e-05,
+      "loss": 0.9629,
+      "step": 134720
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977637972793883e-05,
+      "loss": 0.8952,
+      "step": 134730
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977561807851082e-05,
+      "loss": 1.0404,
+      "step": 134740
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.9774856429082826e-05,
+      "loss": 0.9126,
+      "step": 134750
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977409477965482e-05,
+      "loss": 0.8954,
+      "step": 134760
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977333313022682e-05,
+      "loss": 0.9972,
+      "step": 134770
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977257148079882e-05,
+      "loss": 0.9324,
+      "step": 134780
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.977180983137082e-05,
+      "loss": 0.9506,
+      "step": 134790
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.977104818194281e-05,
+      "loss": 0.8031,
+      "step": 134800
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.977028653251482e-05,
+      "loss": 0.9639,
+      "step": 134810
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.9769524883086815e-05,
+      "loss": 0.8965,
+      "step": 134820
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.976876323365881e-05,
+      "loss": 0.917,
+      "step": 134830
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.976800158423081e-05,
+      "loss": 0.8677,
+      "step": 134840
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.9767239934802814e-05,
+      "loss": 1.016,
+      "step": 134850
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.976647828537481e-05,
+      "loss": 0.8473,
+      "step": 134860
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.976571663594681e-05,
+      "loss": 0.911,
+      "step": 134870
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.976495498651881e-05,
+      "loss": 0.8562,
+      "step": 134880
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.9764193337090804e-05,
+      "loss": 1.0498,
+      "step": 134890
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.976343168766281e-05,
+      "loss": 1.0056,
+      "step": 134900
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.97626700382348e-05,
+      "loss": 0.8854,
+      "step": 134910
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.97619083888068e-05,
+      "loss": 0.842,
+      "step": 134920
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.97611467393788e-05,
+      "loss": 0.9373,
+      "step": 134930
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.97603850899508e-05,
+      "loss": 0.876,
+      "step": 134940
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.9759623440522795e-05,
+      "loss": 1.0074,
+      "step": 134950
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.97588617910948e-05,
+      "loss": 0.8386,
+      "step": 134960
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.975810014166679e-05,
+      "loss": 0.8296,
+      "step": 134970
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.9757338492238795e-05,
+      "loss": 0.9895,
+      "step": 134980
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.975657684281079e-05,
+      "loss": 1.0357,
+      "step": 134990
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.975581519338279e-05,
+      "loss": 0.9823,
+      "step": 135000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.975505354395479e-05,
+      "loss": 0.9441,
+      "step": 135010
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.975429189452679e-05,
+      "loss": 0.9973,
+      "step": 135020
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.975353024509879e-05,
+      "loss": 0.9589,
+      "step": 135030
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9752768595670786e-05,
+      "loss": 0.9243,
+      "step": 135040
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9752006946242784e-05,
+      "loss": 1.0283,
+      "step": 135050
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.975124529681478e-05,
+      "loss": 0.8149,
+      "step": 135060
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9750483647386785e-05,
+      "loss": 0.8888,
+      "step": 135070
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974972199795878e-05,
+      "loss": 0.959,
+      "step": 135080
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974896034853078e-05,
+      "loss": 0.9713,
+      "step": 135090
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974819869910278e-05,
+      "loss": 0.9054,
+      "step": 135100
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974743704967478e-05,
+      "loss": 0.9121,
+      "step": 135110
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974667540024677e-05,
+      "loss": 0.9572,
+      "step": 135120
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974591375081878e-05,
+      "loss": 0.8914,
+      "step": 135130
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9745152101390774e-05,
+      "loss": 0.9918,
+      "step": 135140
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974439045196277e-05,
+      "loss": 0.89,
+      "step": 135150
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974362880253477e-05,
+      "loss": 0.928,
+      "step": 135160
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9742867153106774e-05,
+      "loss": 0.8729,
+      "step": 135170
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9742105503678764e-05,
+      "loss": 0.9561,
+      "step": 135180
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.974134385425077e-05,
+      "loss": 0.8825,
+      "step": 135190
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9740582204822766e-05,
+      "loss": 0.9996,
+      "step": 135200
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9739820555394763e-05,
+      "loss": 0.9051,
+      "step": 135210
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.973905890596676e-05,
+      "loss": 1.0038,
+      "step": 135220
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.9738297256538765e-05,
+      "loss": 0.9171,
+      "step": 135230
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973753560711076e-05,
+      "loss": 0.963,
+      "step": 135240
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973677395768276e-05,
+      "loss": 0.9774,
+      "step": 135250
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973601230825476e-05,
+      "loss": 0.9325,
+      "step": 135260
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.9735250658826755e-05,
+      "loss": 0.931,
+      "step": 135270
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973448900939876e-05,
+      "loss": 0.9305,
+      "step": 135280
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973372735997076e-05,
+      "loss": 1.0041,
+      "step": 135290
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.9732965710542754e-05,
+      "loss": 1.0738,
+      "step": 135300
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973220406111475e-05,
+      "loss": 0.928,
+      "step": 135310
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.9731442411686756e-05,
+      "loss": 1.0232,
+      "step": 135320
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.973068076225875e-05,
+      "loss": 1.0418,
+      "step": 135330
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972991911283075e-05,
+      "loss": 0.9216,
+      "step": 135340
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972915746340274e-05,
+      "loss": 1.0357,
+      "step": 135350
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.9728395813974746e-05,
+      "loss": 0.9608,
+      "step": 135360
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972763416454674e-05,
+      "loss": 0.9237,
+      "step": 135370
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972687251511874e-05,
+      "loss": 0.9337,
+      "step": 135380
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972611086569074e-05,
+      "loss": 0.8948,
+      "step": 135390
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972534921626274e-05,
+      "loss": 0.952,
+      "step": 135400
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972458756683474e-05,
+      "loss": 0.8844,
+      "step": 135410
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972382591740674e-05,
+      "loss": 0.9477,
+      "step": 135420
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.9723064267978735e-05,
+      "loss": 0.9192,
+      "step": 135430
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.972230261855073e-05,
+      "loss": 0.9447,
+      "step": 135440
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9721540969122737e-05,
+      "loss": 0.9869,
+      "step": 135450
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9720779319694734e-05,
+      "loss": 0.8969,
+      "step": 135460
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.972001767026673e-05,
+      "loss": 0.9112,
+      "step": 135470
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971925602083873e-05,
+      "loss": 1.0936,
+      "step": 135480
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971849437141073e-05,
+      "loss": 0.9136,
+      "step": 135490
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9717732721982724e-05,
+      "loss": 0.9208,
+      "step": 135500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971697107255473e-05,
+      "loss": 1.0113,
+      "step": 135510
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9716209423126726e-05,
+      "loss": 1.052,
+      "step": 135520
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971544777369872e-05,
+      "loss": 1.007,
+      "step": 135530
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971468612427072e-05,
+      "loss": 1.0797,
+      "step": 135540
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9713924474842725e-05,
+      "loss": 0.9795,
+      "step": 135550
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9713162825414715e-05,
+      "loss": 0.9033,
+      "step": 135560
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971240117598672e-05,
+      "loss": 0.9675,
+      "step": 135570
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971163952655872e-05,
+      "loss": 0.9081,
+      "step": 135580
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9710877877130715e-05,
+      "loss": 0.9815,
+      "step": 135590
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.971011622770271e-05,
+      "loss": 0.9505,
+      "step": 135600
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9709354578274716e-05,
+      "loss": 0.9564,
+      "step": 135610
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9708592928846714e-05,
+      "loss": 0.969,
+      "step": 135620
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.970783127941871e-05,
+      "loss": 0.9019,
+      "step": 135630
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.970706962999071e-05,
+      "loss": 1.1116,
+      "step": 135640
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.9706307980562706e-05,
+      "loss": 0.8999,
+      "step": 135650
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.970554633113471e-05,
+      "loss": 0.8333,
+      "step": 135660
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.970478468170671e-05,
+      "loss": 0.9036,
+      "step": 135670
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9704023032278705e-05,
+      "loss": 0.9926,
+      "step": 135680
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.97032613828507e-05,
+      "loss": 0.9032,
+      "step": 135690
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.970249973342271e-05,
+      "loss": 1.0019,
+      "step": 135700
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.97017380839947e-05,
+      "loss": 0.9482,
+      "step": 135710
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.97009764345667e-05,
+      "loss": 1.0356,
+      "step": 135720
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.97002147851387e-05,
+      "loss": 0.9238,
+      "step": 135730
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.96994531357107e-05,
+      "loss": 0.9081,
+      "step": 135740
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9698691486282694e-05,
+      "loss": 0.9622,
+      "step": 135750
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.96979298368547e-05,
+      "loss": 0.8243,
+      "step": 135760
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.969716818742669e-05,
+      "loss": 0.9616,
+      "step": 135770
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9696406537998694e-05,
+      "loss": 0.9172,
+      "step": 135780
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.969564488857069e-05,
+      "loss": 0.8838,
+      "step": 135790
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.969488323914269e-05,
+      "loss": 0.9079,
+      "step": 135800
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9694121589714686e-05,
+      "loss": 0.9055,
+      "step": 135810
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9693359940286683e-05,
+      "loss": 1.0813,
+      "step": 135820
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.969259829085869e-05,
+      "loss": 1.1004,
+      "step": 135830
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9691836641430685e-05,
+      "loss": 1.0196,
+      "step": 135840
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.969107499200268e-05,
+      "loss": 0.866,
+      "step": 135850
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.969031334257468e-05,
+      "loss": 0.9374,
+      "step": 135860
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9689551693146684e-05,
+      "loss": 0.8957,
+      "step": 135870
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.9688790043718675e-05,
+      "loss": 0.8815,
+      "step": 135880
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968802839429068e-05,
+      "loss": 0.9697,
+      "step": 135890
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968726674486268e-05,
+      "loss": 0.9695,
+      "step": 135900
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.9686505095434674e-05,
+      "loss": 0.9742,
+      "step": 135910
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968574344600667e-05,
+      "loss": 0.9747,
+      "step": 135920
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.9684981796578676e-05,
+      "loss": 0.9043,
+      "step": 135930
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968422014715067e-05,
+      "loss": 0.9019,
+      "step": 135940
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968345849772267e-05,
+      "loss": 0.8556,
+      "step": 135950
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968269684829467e-05,
+      "loss": 1.0172,
+      "step": 135960
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.9681935198866666e-05,
+      "loss": 0.9838,
+      "step": 135970
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968117354943866e-05,
+      "loss": 0.8667,
+      "step": 135980
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.968041190001067e-05,
+      "loss": 0.9662,
+      "step": 135990
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.9679650250582665e-05,
+      "loss": 0.9378,
+      "step": 136000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967888860115466e-05,
+      "loss": 0.9167,
+      "step": 136010
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967812695172666e-05,
+      "loss": 0.9453,
+      "step": 136020
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967736530229866e-05,
+      "loss": 0.9432,
+      "step": 136030
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967660365287066e-05,
+      "loss": 0.8929,
+      "step": 136040
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967584200344266e-05,
+      "loss": 0.9749,
+      "step": 136050
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.9675080354014657e-05,
+      "loss": 0.91,
+      "step": 136060
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.9674318704586654e-05,
+      "loss": 0.8483,
+      "step": 136070
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967355705515866e-05,
+      "loss": 1.0201,
+      "step": 136080
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967279540573065e-05,
+      "loss": 0.9091,
+      "step": 136090
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.967203375630265e-05,
+      "loss": 0.9707,
+      "step": 136100
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.967127210687465e-05,
+      "loss": 0.8564,
+      "step": 136110
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.967051045744665e-05,
+      "loss": 0.953,
+      "step": 136120
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.9669748808018646e-05,
+      "loss": 0.868,
+      "step": 136130
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966898715859065e-05,
+      "loss": 0.9754,
+      "step": 136140
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966822550916264e-05,
+      "loss": 0.9067,
+      "step": 136150
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.9667463859734645e-05,
+      "loss": 0.9317,
+      "step": 136160
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966670221030664e-05,
+      "loss": 0.9183,
+      "step": 136170
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966594056087864e-05,
+      "loss": 1.0215,
+      "step": 136180
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966517891145064e-05,
+      "loss": 1.1025,
+      "step": 136190
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966441726202264e-05,
+      "loss": 0.8978,
+      "step": 136200
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966365561259464e-05,
+      "loss": 0.9983,
+      "step": 136210
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.9662893963166636e-05,
+      "loss": 0.9279,
+      "step": 136220
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.9662132313738634e-05,
+      "loss": 0.8923,
+      "step": 136230
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.966137066431063e-05,
+      "loss": 0.9823,
+      "step": 136240
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.9660609014882636e-05,
+      "loss": 0.9274,
+      "step": 136250
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.965984736545463e-05,
+      "loss": 0.9394,
+      "step": 136260
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.965908571602663e-05,
+      "loss": 1.0065,
+      "step": 136270
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.965832406659863e-05,
+      "loss": 0.8757,
+      "step": 136280
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.9657562417170625e-05,
+      "loss": 1.0075,
+      "step": 136290
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.965680076774262e-05,
+      "loss": 0.9612,
+      "step": 136300
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.965603911831463e-05,
+      "loss": 0.8578,
+      "step": 136310
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.965527746888662e-05,
+      "loss": 0.9968,
+      "step": 136320
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.965451581945862e-05,
+      "loss": 0.9637,
+      "step": 136330
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.965375417003062e-05,
+      "loss": 0.8849,
+      "step": 136340
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.965299252060262e-05,
+      "loss": 0.977,
+      "step": 136350
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.9652230871174614e-05,
+      "loss": 0.8818,
+      "step": 136360
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.965146922174662e-05,
+      "loss": 0.9047,
+      "step": 136370
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.9650707572318616e-05,
+      "loss": 0.9587,
+      "step": 136380
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.9649945922890614e-05,
+      "loss": 1.0547,
+      "step": 136390
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.964918427346261e-05,
+      "loss": 0.9874,
+      "step": 136400
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.964842262403461e-05,
+      "loss": 1.033,
+      "step": 136410
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.964766097460661e-05,
+      "loss": 0.8902,
+      "step": 136420
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.964689932517861e-05,
+      "loss": 0.8714,
+      "step": 136430
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.964613767575061e-05,
+      "loss": 0.8528,
+      "step": 136440
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.9645376026322605e-05,
+      "loss": 0.9272,
+      "step": 136450
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.964461437689461e-05,
+      "loss": 0.8597,
+      "step": 136460
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.96438527274666e-05,
+      "loss": 0.9349,
+      "step": 136470
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.9643091078038604e-05,
+      "loss": 0.9996,
+      "step": 136480
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.96423294286106e-05,
+      "loss": 0.9965,
+      "step": 136490
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.96415677791826e-05,
+      "loss": 0.9346,
+      "step": 136500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.96408061297546e-05,
+      "loss": 0.9501,
+      "step": 136510
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.96400444803266e-05,
+      "loss": 0.9822,
+      "step": 136520
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.963928283089859e-05,
+      "loss": 0.8934,
+      "step": 136530
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.9638521181470596e-05,
+      "loss": 0.8872,
+      "step": 136540
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.9637759532042593e-05,
+      "loss": 0.9662,
+      "step": 136550
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963699788261459e-05,
+      "loss": 0.9013,
+      "step": 136560
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963623623318659e-05,
+      "loss": 1.1068,
+      "step": 136570
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963547458375859e-05,
+      "loss": 1.0217,
+      "step": 136580
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963471293433059e-05,
+      "loss": 0.9317,
+      "step": 136590
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963395128490259e-05,
+      "loss": 1.0677,
+      "step": 136600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.9633189635474585e-05,
+      "loss": 0.9808,
+      "step": 136610
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963242798604658e-05,
+      "loss": 0.9479,
+      "step": 136620
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963166633661859e-05,
+      "loss": 1.04,
+      "step": 136630
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.9630904687190584e-05,
+      "loss": 0.9879,
+      "step": 136640
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.963014303776258e-05,
+      "loss": 0.9694,
+      "step": 136650
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962938138833458e-05,
+      "loss": 0.8448,
+      "step": 136660
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962861973890658e-05,
+      "loss": 0.9616,
+      "step": 136670
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.9627858089478574e-05,
+      "loss": 0.8986,
+      "step": 136680
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962709644005058e-05,
+      "loss": 0.9935,
+      "step": 136690
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.9626334790622576e-05,
+      "loss": 0.8442,
+      "step": 136700
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962557314119457e-05,
+      "loss": 1.1516,
+      "step": 136710
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962481149176657e-05,
+      "loss": 0.8645,
+      "step": 136720
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962404984233857e-05,
+      "loss": 0.8947,
+      "step": 136730
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.9623288192910566e-05,
+      "loss": 0.8573,
+      "step": 136740
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962252654348257e-05,
+      "loss": 0.9497,
+      "step": 136750
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.962176489405456e-05,
+      "loss": 0.9006,
+      "step": 136760
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9621003244626565e-05,
+      "loss": 1.0129,
+      "step": 136770
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.962024159519856e-05,
+      "loss": 0.9336,
+      "step": 136780
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961947994577056e-05,
+      "loss": 0.8722,
+      "step": 136790
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9618718296342564e-05,
+      "loss": 0.9604,
+      "step": 136800
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961795664691456e-05,
+      "loss": 1.0128,
+      "step": 136810
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961719499748656e-05,
+      "loss": 1.0011,
+      "step": 136820
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9616433348058556e-05,
+      "loss": 1.1064,
+      "step": 136830
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961567169863056e-05,
+      "loss": 0.8171,
+      "step": 136840
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961491004920255e-05,
+      "loss": 1.022,
+      "step": 136850
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9614148399774556e-05,
+      "loss": 0.9001,
+      "step": 136860
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961338675034655e-05,
+      "loss": 0.9144,
+      "step": 136870
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961262510091855e-05,
+      "loss": 0.9072,
+      "step": 136880
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961186345149055e-05,
+      "loss": 1.0565,
+      "step": 136890
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961110180206255e-05,
+      "loss": 0.9761,
+      "step": 136900
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.961034015263454e-05,
+      "loss": 1.0313,
+      "step": 136910
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.960957850320655e-05,
+      "loss": 0.9545,
+      "step": 136920
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9608816853778545e-05,
+      "loss": 1.0274,
+      "step": 136930
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.960805520435054e-05,
+      "loss": 0.8653,
+      "step": 136940
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.960729355492254e-05,
+      "loss": 1.0162,
+      "step": 136950
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9606531905494544e-05,
+      "loss": 0.8959,
+      "step": 136960
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.9605770256066534e-05,
+      "loss": 0.8733,
+      "step": 136970
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.960500860663854e-05,
+      "loss": 0.8899,
+      "step": 136980
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9604246957210536e-05,
+      "loss": 1.0491,
+      "step": 136990
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9603485307782534e-05,
+      "loss": 0.8265,
+      "step": 137000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.960272365835454e-05,
+      "loss": 0.9121,
+      "step": 137010
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9601962008926535e-05,
+      "loss": 0.8928,
+      "step": 137020
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.960120035949853e-05,
+      "loss": 0.9224,
+      "step": 137030
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.960043871007053e-05,
+      "loss": 0.9107,
+      "step": 137040
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9599677060642535e-05,
+      "loss": 0.9762,
+      "step": 137050
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9598915411214525e-05,
+      "loss": 0.8811,
+      "step": 137060
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959815376178653e-05,
+      "loss": 0.8671,
+      "step": 137070
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959739211235853e-05,
+      "loss": 0.9114,
+      "step": 137080
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9596630462930524e-05,
+      "loss": 0.9562,
+      "step": 137090
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959586881350252e-05,
+      "loss": 0.9406,
+      "step": 137100
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9595107164074526e-05,
+      "loss": 0.8461,
+      "step": 137110
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959434551464652e-05,
+      "loss": 0.994,
+      "step": 137120
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959358386521852e-05,
+      "loss": 0.98,
+      "step": 137130
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959282221579052e-05,
+      "loss": 0.8873,
+      "step": 137140
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9592060566362516e-05,
+      "loss": 0.9989,
+      "step": 137150
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9591298916934513e-05,
+      "loss": 0.9799,
+      "step": 137160
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.959053726750652e-05,
+      "loss": 0.8738,
+      "step": 137170
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.9589775618078515e-05,
+      "loss": 0.9804,
+      "step": 137180
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.958901396865051e-05,
+      "loss": 0.9764,
+      "step": 137190
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.958825231922251e-05,
+      "loss": 0.8708,
+      "step": 137200
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.958749066979451e-05,
+      "loss": 0.9417,
+      "step": 137210
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.958672902036651e-05,
+      "loss": 0.9262,
+      "step": 137220
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.95859673709385e-05,
+      "loss": 0.9156,
+      "step": 137230
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.958520572151051e-05,
+      "loss": 0.8455,
+      "step": 137240
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9584444072082504e-05,
+      "loss": 1.0944,
+      "step": 137250
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.95836824226545e-05,
+      "loss": 1.0387,
+      "step": 137260
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.95829207732265e-05,
+      "loss": 0.831,
+      "step": 137270
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.95821591237985e-05,
+      "loss": 0.8815,
+      "step": 137280
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9581397474370494e-05,
+      "loss": 0.9516,
+      "step": 137290
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.95806358249425e-05,
+      "loss": 0.9745,
+      "step": 137300
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9579874175514496e-05,
+      "loss": 0.9054,
+      "step": 137310
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.957911252608649e-05,
+      "loss": 0.9205,
+      "step": 137320
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.957835087665849e-05,
+      "loss": 0.8869,
+      "step": 137330
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9577589227230495e-05,
+      "loss": 0.8816,
+      "step": 137340
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9576827577802486e-05,
+      "loss": 0.9483,
+      "step": 137350
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.957606592837449e-05,
+      "loss": 1.0432,
+      "step": 137360
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.957530427894649e-05,
+      "loss": 0.9457,
+      "step": 137370
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9574542629518485e-05,
+      "loss": 0.8997,
+      "step": 137380
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.957378098009049e-05,
+      "loss": 1.1056,
+      "step": 137390
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9573019330662487e-05,
+      "loss": 0.9911,
+      "step": 137400
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.9572257681234484e-05,
+      "loss": 0.9815,
+      "step": 137410
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.957149603180648e-05,
+      "loss": 0.9077,
+      "step": 137420
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.9570734382378486e-05,
+      "loss": 0.9926,
+      "step": 137430
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.9569972732950476e-05,
+      "loss": 1.0152,
+      "step": 137440
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956921108352248e-05,
+      "loss": 0.9441,
+      "step": 137450
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956844943409448e-05,
+      "loss": 0.942,
+      "step": 137460
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.9567687784666476e-05,
+      "loss": 0.9853,
+      "step": 137470
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956692613523847e-05,
+      "loss": 0.9327,
+      "step": 137480
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956616448581048e-05,
+      "loss": 0.8922,
+      "step": 137490
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956540283638247e-05,
+      "loss": 0.8173,
+      "step": 137500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956464118695447e-05,
+      "loss": 0.9628,
+      "step": 137510
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956387953752647e-05,
+      "loss": 0.9457,
+      "step": 137520
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956311788809847e-05,
+      "loss": 0.969,
+      "step": 137530
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.9562356238670465e-05,
+      "loss": 0.8774,
+      "step": 137540
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956159458924247e-05,
+      "loss": 0.9689,
+      "step": 137550
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.956083293981446e-05,
+      "loss": 1.0248,
+      "step": 137560
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.9560071290386464e-05,
+      "loss": 1.0269,
+      "step": 137570
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.955930964095846e-05,
+      "loss": 0.9545,
+      "step": 137580
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.955854799153046e-05,
+      "loss": 0.8945,
+      "step": 137590
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.955778634210246e-05,
+      "loss": 1.0143,
+      "step": 137600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.955702469267446e-05,
+      "loss": 0.9136,
+      "step": 137610
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.955626304324646e-05,
+      "loss": 0.8438,
+      "step": 137620
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.9555501393818455e-05,
+      "loss": 0.8554,
+      "step": 137630
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.955473974439046e-05,
+      "loss": 0.9289,
+      "step": 137640
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.955397809496245e-05,
+      "loss": 0.9065,
+      "step": 137650
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9553216445534455e-05,
+      "loss": 1.0198,
+      "step": 137660
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9552454796106445e-05,
+      "loss": 0.9281,
+      "step": 137670
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.955169314667845e-05,
+      "loss": 0.8574,
+      "step": 137680
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.955093149725045e-05,
+      "loss": 1.0529,
+      "step": 137690
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9550169847822444e-05,
+      "loss": 0.9367,
+      "step": 137700
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954940819839444e-05,
+      "loss": 0.8634,
+      "step": 137710
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9548646548966446e-05,
+      "loss": 1.0193,
+      "step": 137720
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954788489953844e-05,
+      "loss": 0.9788,
+      "step": 137730
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954712325011044e-05,
+      "loss": 1.0177,
+      "step": 137740
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954636160068244e-05,
+      "loss": 0.9017,
+      "step": 137750
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9545599951254436e-05,
+      "loss": 0.9907,
+      "step": 137760
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9544838301826433e-05,
+      "loss": 0.9583,
+      "step": 137770
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954407665239844e-05,
+      "loss": 0.802,
+      "step": 137780
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.9543315002970435e-05,
+      "loss": 0.9118,
+      "step": 137790
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954255335354243e-05,
+      "loss": 0.8282,
+      "step": 137800
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954179170411444e-05,
+      "loss": 0.86,
+      "step": 137810
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954103005468643e-05,
+      "loss": 1.0718,
+      "step": 137820
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.954026840525843e-05,
+      "loss": 0.9901,
+      "step": 137830
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.953950675583043e-05,
+      "loss": 0.816,
+      "step": 137840
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.953874510640243e-05,
+      "loss": 0.9832,
+      "step": 137850
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.9537983456974424e-05,
+      "loss": 0.8978,
+      "step": 137860
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953722180754643e-05,
+      "loss": 0.9029,
+      "step": 137870
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953646015811842e-05,
+      "loss": 0.8755,
+      "step": 137880
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953569850869042e-05,
+      "loss": 0.9235,
+      "step": 137890
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953493685926242e-05,
+      "loss": 1.0702,
+      "step": 137900
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953417520983442e-05,
+      "loss": 1.0021,
+      "step": 137910
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.9533413560406416e-05,
+      "loss": 0.9413,
+      "step": 137920
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953265191097842e-05,
+      "loss": 0.8135,
+      "step": 137930
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953189026155041e-05,
+      "loss": 0.9953,
+      "step": 137940
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.9531128612122415e-05,
+      "loss": 0.8953,
+      "step": 137950
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.953036696269441e-05,
+      "loss": 0.9975,
+      "step": 137960
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.952960531326641e-05,
+      "loss": 0.9524,
+      "step": 137970
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.9528843663838414e-05,
+      "loss": 1.1514,
+      "step": 137980
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.952808201441041e-05,
+      "loss": 0.927,
+      "step": 137990
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.952732036498241e-05,
+      "loss": 0.9733,
+      "step": 138000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.9526558715554407e-05,
+      "loss": 0.9605,
+      "step": 138010
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.952579706612641e-05,
+      "loss": 0.9432,
+      "step": 138020
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.95250354166984e-05,
+      "loss": 0.9689,
+      "step": 138030
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.9524273767270406e-05,
+      "loss": 0.8809,
+      "step": 138040
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.95235121178424e-05,
+      "loss": 0.9842,
+      "step": 138050
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.95227504684144e-05,
+      "loss": 0.8506,
+      "step": 138060
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.95219888189864e-05,
+      "loss": 0.9872,
+      "step": 138070
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.95212271695584e-05,
+      "loss": 0.92,
+      "step": 138080
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.952046552013039e-05,
+      "loss": 1.0582,
+      "step": 138090
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.95197038707024e-05,
+      "loss": 1.0804,
+      "step": 138100
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951894222127439e-05,
+      "loss": 0.9823,
+      "step": 138110
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951818057184639e-05,
+      "loss": 0.9641,
+      "step": 138120
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951741892241839e-05,
+      "loss": 0.8893,
+      "step": 138130
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951665727299039e-05,
+      "loss": 0.8919,
+      "step": 138140
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.9515895623562385e-05,
+      "loss": 0.9627,
+      "step": 138150
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951513397413439e-05,
+      "loss": 0.9861,
+      "step": 138160
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.9514372324706386e-05,
+      "loss": 0.9282,
+      "step": 138170
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.9513610675278384e-05,
+      "loss": 0.8539,
+      "step": 138180
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951284902585039e-05,
+      "loss": 0.9103,
+      "step": 138190
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951208737642238e-05,
+      "loss": 0.8941,
+      "step": 138200
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951132572699438e-05,
+      "loss": 0.7509,
+      "step": 138210
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.951056407756638e-05,
+      "loss": 1.0061,
+      "step": 138220
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.950980242813838e-05,
+      "loss": 1.0164,
+      "step": 138230
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.9509040778710375e-05,
+      "loss": 0.9655,
+      "step": 138240
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.950827912928238e-05,
+      "loss": 0.955,
+      "step": 138250
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.950751747985437e-05,
+      "loss": 1.0015,
+      "step": 138260
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.9506755830426375e-05,
+      "loss": 1.0016,
+      "step": 138270
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.950599418099837e-05,
+      "loss": 1.0227,
+      "step": 138280
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.950523253157037e-05,
+      "loss": 0.8877,
+      "step": 138290
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.950447088214237e-05,
+      "loss": 0.9328,
+      "step": 138300
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.950370923271437e-05,
+      "loss": 0.887,
+      "step": 138310
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.950294758328636e-05,
+      "loss": 0.8205,
+      "step": 138320
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.9502185933858366e-05,
+      "loss": 1.1522,
+      "step": 138330
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.9501424284430364e-05,
+      "loss": 1.0264,
+      "step": 138340
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.950066263500236e-05,
+      "loss": 0.9596,
+      "step": 138350
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949990098557436e-05,
+      "loss": 0.9165,
+      "step": 138360
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949913933614636e-05,
+      "loss": 0.9517,
+      "step": 138370
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949837768671836e-05,
+      "loss": 0.9675,
+      "step": 138380
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949761603729036e-05,
+      "loss": 0.9642,
+      "step": 138390
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949685438786236e-05,
+      "loss": 0.9059,
+      "step": 138400
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949609273843435e-05,
+      "loss": 0.8788,
+      "step": 138410
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949533108900636e-05,
+      "loss": 1.0993,
+      "step": 138420
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.9494569439578354e-05,
+      "loss": 1.0366,
+      "step": 138430
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949380779015035e-05,
+      "loss": 0.8982,
+      "step": 138440
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949304614072235e-05,
+      "loss": 1.0194,
+      "step": 138450
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.9492284491294354e-05,
+      "loss": 0.9328,
+      "step": 138460
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.9491522841866344e-05,
+      "loss": 0.9497,
+      "step": 138470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.949076119243835e-05,
+      "loss": 0.9326,
+      "step": 138480
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.9489999543010346e-05,
+      "loss": 1.0755,
+      "step": 138490
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.948923789358234e-05,
+      "loss": 0.9683,
+      "step": 138500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.948847624415434e-05,
+      "loss": 1.0114,
+      "step": 138510
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9487714594726345e-05,
+      "loss": 1.017,
+      "step": 138520
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9486952945298336e-05,
+      "loss": 0.9554,
+      "step": 138530
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.948619129587034e-05,
+      "loss": 0.8859,
+      "step": 138540
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.948542964644234e-05,
+      "loss": 0.9261,
+      "step": 138550
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9484667997014335e-05,
+      "loss": 1.0801,
+      "step": 138560
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.948390634758633e-05,
+      "loss": 0.9443,
+      "step": 138570
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.948314469815833e-05,
+      "loss": 0.8518,
+      "step": 138580
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9482383048730334e-05,
+      "loss": 0.9338,
+      "step": 138590
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.948162139930233e-05,
+      "loss": 0.8459,
+      "step": 138600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.948085974987433e-05,
+      "loss": 0.8876,
+      "step": 138610
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9480098100446327e-05,
+      "loss": 0.8696,
+      "step": 138620
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947933645101833e-05,
+      "loss": 0.9373,
+      "step": 138630
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947857480159032e-05,
+      "loss": 1.0713,
+      "step": 138640
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9477813152162326e-05,
+      "loss": 1.003,
+      "step": 138650
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947705150273432e-05,
+      "loss": 0.9798,
+      "step": 138660
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947628985330632e-05,
+      "loss": 0.9465,
+      "step": 138670
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947552820387832e-05,
+      "loss": 0.9719,
+      "step": 138680
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947476655445032e-05,
+      "loss": 0.9719,
+      "step": 138690
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947400490502231e-05,
+      "loss": 0.7789,
+      "step": 138700
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947324325559432e-05,
+      "loss": 0.9449,
+      "step": 138710
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.9472481606166315e-05,
+      "loss": 0.9707,
+      "step": 138720
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.947171995673831e-05,
+      "loss": 0.9406,
+      "step": 138730
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.947095830731031e-05,
+      "loss": 0.975,
+      "step": 138740
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9470196657882314e-05,
+      "loss": 0.8873,
+      "step": 138750
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.946943500845431e-05,
+      "loss": 0.8526,
+      "step": 138760
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.946867335902631e-05,
+      "loss": 0.8676,
+      "step": 138770
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.946791170959831e-05,
+      "loss": 1.0108,
+      "step": 138780
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9467150060170304e-05,
+      "loss": 0.9707,
+      "step": 138790
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.946638841074231e-05,
+      "loss": 0.9263,
+      "step": 138800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9465626761314306e-05,
+      "loss": 0.9765,
+      "step": 138810
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.94648651118863e-05,
+      "loss": 0.9373,
+      "step": 138820
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.94641034624583e-05,
+      "loss": 1.1179,
+      "step": 138830
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9463341813030305e-05,
+      "loss": 0.8499,
+      "step": 138840
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9462580163602295e-05,
+      "loss": 0.9193,
+      "step": 138850
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.94618185141743e-05,
+      "loss": 0.9855,
+      "step": 138860
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.94610568647463e-05,
+      "loss": 1.0231,
+      "step": 138870
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9460295215318295e-05,
+      "loss": 1.0094,
+      "step": 138880
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.945953356589029e-05,
+      "loss": 0.961,
+      "step": 138890
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9458771916462296e-05,
+      "loss": 1.1193,
+      "step": 138900
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.945801026703429e-05,
+      "loss": 0.9667,
+      "step": 138910
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.945724861760629e-05,
+      "loss": 0.9686,
+      "step": 138920
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.945648696817829e-05,
+      "loss": 0.9714,
+      "step": 138930
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9455725318750286e-05,
+      "loss": 0.9906,
+      "step": 138940
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9454963669322284e-05,
+      "loss": 1.0147,
+      "step": 138950
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.945420201989429e-05,
+      "loss": 1.0074,
+      "step": 138960
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9453440370466285e-05,
+      "loss": 0.9773,
+      "step": 138970
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.945267872103828e-05,
+      "loss": 0.9015,
+      "step": 138980
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.945191707161029e-05,
+      "loss": 0.9807,
+      "step": 138990
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.945115542218228e-05,
+      "loss": 1.0263,
+      "step": 139000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.945039377275428e-05,
+      "loss": 0.9512,
+      "step": 139010
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944963212332628e-05,
+      "loss": 0.9547,
+      "step": 139020
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944887047389828e-05,
+      "loss": 1.0285,
+      "step": 139030
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9448108824470274e-05,
+      "loss": 1.0433,
+      "step": 139040
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944734717504227e-05,
+      "loss": 0.9168,
+      "step": 139050
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944658552561427e-05,
+      "loss": 0.9707,
+      "step": 139060
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9445823876186274e-05,
+      "loss": 0.85,
+      "step": 139070
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9445062226758264e-05,
+      "loss": 0.9394,
+      "step": 139080
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944430057733027e-05,
+      "loss": 0.977,
+      "step": 139090
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9443538927902266e-05,
+      "loss": 0.8763,
+      "step": 139100
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9442777278474263e-05,
+      "loss": 1.0036,
+      "step": 139110
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944201562904626e-05,
+      "loss": 0.9139,
+      "step": 139120
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9441253979618265e-05,
+      "loss": 0.9939,
+      "step": 139130
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.944049233019026e-05,
+      "loss": 0.9318,
+      "step": 139140
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.943973068076226e-05,
+      "loss": 0.9283,
+      "step": 139150
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.943896903133426e-05,
+      "loss": 0.9102,
+      "step": 139160
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9438207381906255e-05,
+      "loss": 0.9561,
+      "step": 139170
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.943744573247826e-05,
+      "loss": 0.89,
+      "step": 139180
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.943668408305026e-05,
+      "loss": 1.0052,
+      "step": 139190
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9435922433622254e-05,
+      "loss": 0.8671,
+      "step": 139200
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.943516078419425e-05,
+      "loss": 0.9183,
+      "step": 139210
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9434399134766256e-05,
+      "loss": 0.9776,
+      "step": 139220
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9433637485338247e-05,
+      "loss": 0.9442,
+      "step": 139230
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.943287583591025e-05,
+      "loss": 0.8512,
+      "step": 139240
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.943211418648225e-05,
+      "loss": 0.9825,
+      "step": 139250
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9431352537054246e-05,
+      "loss": 1.0017,
+      "step": 139260
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.943059088762624e-05,
+      "loss": 0.9941,
+      "step": 139270
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942982923819825e-05,
+      "loss": 0.7892,
+      "step": 139280
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942906758877024e-05,
+      "loss": 0.9421,
+      "step": 139290
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942830593934224e-05,
+      "loss": 0.9235,
+      "step": 139300
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942754428991424e-05,
+      "loss": 0.9269,
+      "step": 139310
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942678264048624e-05,
+      "loss": 0.9885,
+      "step": 139320
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9426020991058235e-05,
+      "loss": 0.8566,
+      "step": 139330
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942525934163024e-05,
+      "loss": 0.926,
+      "step": 139340
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9424497692202236e-05,
+      "loss": 0.9737,
+      "step": 139350
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9423736042774234e-05,
+      "loss": 0.8832,
+      "step": 139360
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942297439334623e-05,
+      "loss": 0.9991,
+      "step": 139370
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942221274391823e-05,
+      "loss": 0.942,
+      "step": 139380
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.942145109449023e-05,
+      "loss": 0.9437,
+      "step": 139390
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.942068944506223e-05,
+      "loss": 0.8979,
+      "step": 139400
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941992779563423e-05,
+      "loss": 0.955,
+      "step": 139410
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9419166146206226e-05,
+      "loss": 0.8549,
+      "step": 139420
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941840449677823e-05,
+      "loss": 1.0101,
+      "step": 139430
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941764284735022e-05,
+      "loss": 0.9568,
+      "step": 139440
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9416881197922225e-05,
+      "loss": 0.9109,
+      "step": 139450
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941611954849422e-05,
+      "loss": 0.9855,
+      "step": 139460
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941535789906622e-05,
+      "loss": 0.9316,
+      "step": 139470
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941459624963822e-05,
+      "loss": 0.9236,
+      "step": 139480
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941383460021022e-05,
+      "loss": 0.9554,
+      "step": 139490
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941307295078221e-05,
+      "loss": 0.9577,
+      "step": 139500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9412311301354216e-05,
+      "loss": 0.867,
+      "step": 139510
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9411549651926214e-05,
+      "loss": 0.887,
+      "step": 139520
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941078800249821e-05,
+      "loss": 0.9296,
+      "step": 139530
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.941002635307021e-05,
+      "loss": 0.8752,
+      "step": 139540
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9409264703642206e-05,
+      "loss": 0.9429,
+      "step": 139550
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.940850305421421e-05,
+      "loss": 0.8924,
+      "step": 139560
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.940774140478621e-05,
+      "loss": 0.8609,
+      "step": 139570
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9406979755358205e-05,
+      "loss": 0.9142,
+      "step": 139580
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.94062181059302e-05,
+      "loss": 0.9158,
+      "step": 139590
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.940545645650221e-05,
+      "loss": 0.9723,
+      "step": 139600
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.94046948070742e-05,
+      "loss": 0.9876,
+      "step": 139610
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.94039331576462e-05,
+      "loss": 0.8933,
+      "step": 139620
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.94031715082182e-05,
+      "loss": 0.9137,
+      "step": 139630
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.94024098587902e-05,
+      "loss": 1.0093,
+      "step": 139640
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.9401648209362194e-05,
+      "loss": 0.9285,
+      "step": 139650
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.94008865599342e-05,
+      "loss": 0.9943,
+      "step": 139660
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.940012491050619e-05,
+      "loss": 0.9824,
+      "step": 139670
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.9399363261078194e-05,
+      "loss": 0.8988,
+      "step": 139680
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939860161165019e-05,
+      "loss": 0.9417,
+      "step": 139690
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939783996222219e-05,
+      "loss": 0.8726,
+      "step": 139700
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.9397078312794186e-05,
+      "loss": 0.8754,
+      "step": 139710
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939631666336619e-05,
+      "loss": 0.9016,
+      "step": 139720
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939555501393819e-05,
+      "loss": 0.9476,
+      "step": 139730
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.9394793364510185e-05,
+      "loss": 0.9155,
+      "step": 139740
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939403171508218e-05,
+      "loss": 0.8785,
+      "step": 139750
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939327006565418e-05,
+      "loss": 0.8852,
+      "step": 139760
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.9392508416226184e-05,
+      "loss": 0.8794,
+      "step": 139770
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939174676679818e-05,
+      "loss": 0.9967,
+      "step": 139780
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939098511737018e-05,
+      "loss": 0.8749,
+      "step": 139790
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.939022346794218e-05,
+      "loss": 0.9452,
+      "step": 139800
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.938946181851418e-05,
+      "loss": 0.9823,
+      "step": 139810
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.938870016908617e-05,
+      "loss": 0.9163,
+      "step": 139820
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9387938519658176e-05,
+      "loss": 1.0123,
+      "step": 139830
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938717687023017e-05,
+      "loss": 0.9629,
+      "step": 139840
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938641522080217e-05,
+      "loss": 0.9386,
+      "step": 139850
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938565357137417e-05,
+      "loss": 0.9204,
+      "step": 139860
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938489192194617e-05,
+      "loss": 1.0148,
+      "step": 139870
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938413027251816e-05,
+      "loss": 0.8308,
+      "step": 139880
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938336862309017e-05,
+      "loss": 0.8311,
+      "step": 139890
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9382606973662165e-05,
+      "loss": 0.9457,
+      "step": 139900
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938184532423416e-05,
+      "loss": 0.8161,
+      "step": 139910
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.938108367480616e-05,
+      "loss": 1.1614,
+      "step": 139920
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9380322025378164e-05,
+      "loss": 0.9728,
+      "step": 139930
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937956037595016e-05,
+      "loss": 0.9115,
+      "step": 139940
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937879872652216e-05,
+      "loss": 0.93,
+      "step": 139950
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9378037077094156e-05,
+      "loss": 1.0337,
+      "step": 139960
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9377275427666154e-05,
+      "loss": 0.9345,
+      "step": 139970
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937651377823816e-05,
+      "loss": 0.9268,
+      "step": 139980
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937575212881015e-05,
+      "loss": 0.9368,
+      "step": 139990
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937499047938215e-05,
+      "loss": 1.0222,
+      "step": 140000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937422882995415e-05,
+      "loss": 0.8155,
+      "step": 140010
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937346718052615e-05,
+      "loss": 0.8921,
+      "step": 140020
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9372705531098146e-05,
+      "loss": 1.065,
+      "step": 140030
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.937194388167015e-05,
+      "loss": 0.9408,
+      "step": 140040
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.937118223224214e-05,
+      "loss": 0.9217,
+      "step": 140050
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9370420582814145e-05,
+      "loss": 0.9166,
+      "step": 140060
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936965893338614e-05,
+      "loss": 0.9425,
+      "step": 140070
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936889728395814e-05,
+      "loss": 0.8509,
+      "step": 140080
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936813563453014e-05,
+      "loss": 1.0377,
+      "step": 140090
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936737398510214e-05,
+      "loss": 0.9949,
+      "step": 140100
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936661233567414e-05,
+      "loss": 1.0776,
+      "step": 140110
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9365850686246136e-05,
+      "loss": 0.8769,
+      "step": 140120
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9365089036818134e-05,
+      "loss": 1.0022,
+      "step": 140130
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936432738739013e-05,
+      "loss": 0.9307,
+      "step": 140140
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9363565737962135e-05,
+      "loss": 0.9996,
+      "step": 140150
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936280408853413e-05,
+      "loss": 0.9422,
+      "step": 140160
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936204243910613e-05,
+      "loss": 0.9583,
+      "step": 140170
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936128078967813e-05,
+      "loss": 0.884,
+      "step": 140180
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.936051914025013e-05,
+      "loss": 0.8942,
+      "step": 140190
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.935975749082212e-05,
+      "loss": 0.9667,
+      "step": 140200
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.935899584139413e-05,
+      "loss": 0.8898,
+      "step": 140210
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9358234191966124e-05,
+      "loss": 0.9296,
+      "step": 140220
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.935747254253812e-05,
+      "loss": 0.9666,
+      "step": 140230
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.935671089311012e-05,
+      "loss": 1.0206,
+      "step": 140240
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9355949243682124e-05,
+      "loss": 0.9925,
+      "step": 140250
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9355187594254114e-05,
+      "loss": 0.9864,
+      "step": 140260
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.935442594482612e-05,
+      "loss": 0.9303,
+      "step": 140270
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9353664295398116e-05,
+      "loss": 0.9028,
+      "step": 140280
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9352902645970114e-05,
+      "loss": 0.9516,
+      "step": 140290
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.935214099654211e-05,
+      "loss": 1.0482,
+      "step": 140300
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9351379347114115e-05,
+      "loss": 0.9538,
+      "step": 140310
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.935061769768611e-05,
+      "loss": 0.8722,
+      "step": 140320
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.934985604825811e-05,
+      "loss": 1.0108,
+      "step": 140330
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.934909439883011e-05,
+      "loss": 0.9629,
+      "step": 140340
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9348332749402105e-05,
+      "loss": 0.9743,
+      "step": 140350
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.934757109997411e-05,
+      "loss": 0.8491,
+      "step": 140360
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.934680945054611e-05,
+      "loss": 0.9429,
+      "step": 140370
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9346047801118104e-05,
+      "loss": 0.9343,
+      "step": 140380
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.93452861516901e-05,
+      "loss": 1.0752,
+      "step": 140390
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9344524502262106e-05,
+      "loss": 0.9369,
+      "step": 140400
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.93437628528341e-05,
+      "loss": 0.972,
+      "step": 140410
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.93430012034061e-05,
+      "loss": 0.954,
+      "step": 140420
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.934223955397809e-05,
+      "loss": 0.9521,
+      "step": 140430
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9341477904550096e-05,
+      "loss": 0.8522,
+      "step": 140440
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.934071625512209e-05,
+      "loss": 0.9165,
+      "step": 140450
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.933995460569409e-05,
+      "loss": 0.9084,
+      "step": 140460
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.933919295626609e-05,
+      "loss": 1.0285,
+      "step": 140470
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.933843130683809e-05,
+      "loss": 0.889,
+      "step": 140480
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933766965741009e-05,
+      "loss": 0.8472,
+      "step": 140490
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933690800798209e-05,
+      "loss": 0.9835,
+      "step": 140500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9336146358554085e-05,
+      "loss": 1.0201,
+      "step": 140510
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933538470912608e-05,
+      "loss": 0.9022,
+      "step": 140520
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933462305969809e-05,
+      "loss": 0.9205,
+      "step": 140530
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9333861410270084e-05,
+      "loss": 0.9849,
+      "step": 140540
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933309976084208e-05,
+      "loss": 0.8611,
+      "step": 140550
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933233811141408e-05,
+      "loss": 1.0049,
+      "step": 140560
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933157646198608e-05,
+      "loss": 0.8594,
+      "step": 140570
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9330814812558074e-05,
+      "loss": 0.9949,
+      "step": 140580
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.933005316313008e-05,
+      "loss": 0.9578,
+      "step": 140590
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9329291513702076e-05,
+      "loss": 0.9513,
+      "step": 140600
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.932852986427407e-05,
+      "loss": 0.9882,
+      "step": 140610
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.932776821484607e-05,
+      "loss": 1.002,
+      "step": 140620
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9327006565418075e-05,
+      "loss": 0.8564,
+      "step": 140630
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9326244915990066e-05,
+      "loss": 0.8929,
+      "step": 140640
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.932548326656207e-05,
+      "loss": 0.9189,
+      "step": 140650
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.932472161713407e-05,
+      "loss": 1.0781,
+      "step": 140660
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9323959967706065e-05,
+      "loss": 0.9811,
+      "step": 140670
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.932319831827806e-05,
+      "loss": 0.874,
+      "step": 140680
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9322436668850066e-05,
+      "loss": 0.8986,
+      "step": 140690
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.9321675019422064e-05,
+      "loss": 1.0248,
+      "step": 140700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.932091336999406e-05,
+      "loss": 0.8955,
+      "step": 140710
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.932015172056606e-05,
+      "loss": 1.1456,
+      "step": 140720
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9319390071138056e-05,
+      "loss": 0.9166,
+      "step": 140730
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931862842171006e-05,
+      "loss": 1.0268,
+      "step": 140740
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931786677228206e-05,
+      "loss": 0.9965,
+      "step": 140750
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9317105122854055e-05,
+      "loss": 0.9289,
+      "step": 140760
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931634347342605e-05,
+      "loss": 1.0516,
+      "step": 140770
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931558182399806e-05,
+      "loss": 0.9759,
+      "step": 140780
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931482017457005e-05,
+      "loss": 0.9877,
+      "step": 140790
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931405852514205e-05,
+      "loss": 1.0469,
+      "step": 140800
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931329687571405e-05,
+      "loss": 0.8619,
+      "step": 140810
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931253522628605e-05,
+      "loss": 0.844,
+      "step": 140820
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9311773576858045e-05,
+      "loss": 0.9456,
+      "step": 140830
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931101192743005e-05,
+      "loss": 0.8901,
+      "step": 140840
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.931025027800204e-05,
+      "loss": 1.0272,
+      "step": 140850
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9309488628574044e-05,
+      "loss": 0.9366,
+      "step": 140860
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.930872697914604e-05,
+      "loss": 0.9229,
+      "step": 140870
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.930796532971804e-05,
+      "loss": 0.8974,
+      "step": 140880
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9307203680290036e-05,
+      "loss": 0.9707,
+      "step": 140890
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9306442030862034e-05,
+      "loss": 0.8859,
+      "step": 140900
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.930568038143404e-05,
+      "loss": 0.9474,
+      "step": 140910
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9304918732006035e-05,
+      "loss": 1.0199,
+      "step": 140920
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.930415708257803e-05,
+      "loss": 0.9221,
+      "step": 140930
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.930339543315003e-05,
+      "loss": 0.8268,
+      "step": 140940
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9302633783722034e-05,
+      "loss": 0.8434,
+      "step": 140950
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9301872134294025e-05,
+      "loss": 0.9392,
+      "step": 140960
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.930111048486603e-05,
+      "loss": 0.9211,
+      "step": 140970
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.930034883543803e-05,
+      "loss": 0.9603,
+      "step": 140980
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9299587186010024e-05,
+      "loss": 0.8712,
+      "step": 140990
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929882553658202e-05,
+      "loss": 0.9252,
+      "step": 141000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9298063887154026e-05,
+      "loss": 0.9499,
+      "step": 141010
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929730223772602e-05,
+      "loss": 0.8987,
+      "step": 141020
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929654058829802e-05,
+      "loss": 0.9263,
+      "step": 141030
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929577893887002e-05,
+      "loss": 0.9421,
+      "step": 141040
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9295017289442016e-05,
+      "loss": 0.9711,
+      "step": 141050
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929425564001401e-05,
+      "loss": 0.9405,
+      "step": 141060
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929349399058602e-05,
+      "loss": 0.9108,
+      "step": 141070
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9292732341158015e-05,
+      "loss": 0.898,
+      "step": 141080
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929197069173001e-05,
+      "loss": 0.9255,
+      "step": 141090
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929120904230201e-05,
+      "loss": 0.9147,
+      "step": 141100
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.929044739287401e-05,
+      "loss": 0.8595,
+      "step": 141110
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.928968574344601e-05,
+      "loss": 0.9332,
+      "step": 141120
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.928892409401801e-05,
+      "loss": 0.9286,
+      "step": 141130
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.928816244459001e-05,
+      "loss": 0.8862,
+      "step": 141140
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9287400795162004e-05,
+      "loss": 0.9219,
+      "step": 141150
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.928663914573401e-05,
+      "loss": 0.9226,
+      "step": 141160
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9285877496306e-05,
+      "loss": 0.9498,
+      "step": 141170
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9285115846878e-05,
+      "loss": 0.9223,
+      "step": 141180
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.928435419745e-05,
+      "loss": 0.9748,
+      "step": 141190
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9283592548022e-05,
+      "loss": 0.7815,
+      "step": 141200
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9282830898593996e-05,
+      "loss": 0.8261,
+      "step": 141210
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9282069249166e-05,
+      "loss": 1.0663,
+      "step": 141220
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.928130759973799e-05,
+      "loss": 0.8278,
+      "step": 141230
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9280545950309995e-05,
+      "loss": 0.9891,
+      "step": 141240
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927978430088199e-05,
+      "loss": 0.8747,
+      "step": 141250
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927902265145399e-05,
+      "loss": 1.0159,
+      "step": 141260
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927826100202599e-05,
+      "loss": 0.9962,
+      "step": 141270
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927749935259799e-05,
+      "loss": 0.9019,
+      "step": 141280
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927673770316999e-05,
+      "loss": 0.9667,
+      "step": 141290
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9275976053741986e-05,
+      "loss": 0.9659,
+      "step": 141300
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9275214404313984e-05,
+      "loss": 0.9198,
+      "step": 141310
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927445275488598e-05,
+      "loss": 1.0138,
+      "step": 141320
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9273691105457986e-05,
+      "loss": 0.9341,
+      "step": 141330
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9272929456029976e-05,
+      "loss": 0.8971,
+      "step": 141340
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927216780660198e-05,
+      "loss": 0.8701,
+      "step": 141350
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.927140615717398e-05,
+      "loss": 0.9412,
+      "step": 141360
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9270644507745975e-05,
+      "loss": 0.8537,
+      "step": 141370
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926988285831797e-05,
+      "loss": 0.9472,
+      "step": 141380
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926912120888998e-05,
+      "loss": 0.9804,
+      "step": 141390
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926835955946197e-05,
+      "loss": 1.0163,
+      "step": 141400
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926759791003397e-05,
+      "loss": 0.9087,
+      "step": 141410
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926683626060597e-05,
+      "loss": 0.984,
+      "step": 141420
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926607461117797e-05,
+      "loss": 0.9861,
+      "step": 141430
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9265312961749965e-05,
+      "loss": 0.9276,
+      "step": 141440
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926455131232197e-05,
+      "loss": 0.9188,
+      "step": 141450
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926378966289396e-05,
+      "loss": 0.9349,
+      "step": 141460
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9263028013465964e-05,
+      "loss": 0.961,
+      "step": 141470
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926226636403796e-05,
+      "loss": 0.9638,
+      "step": 141480
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926150471460996e-05,
+      "loss": 0.942,
+      "step": 141490
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.926074306518196e-05,
+      "loss": 0.9643,
+      "step": 141500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.925998141575396e-05,
+      "loss": 0.909,
+      "step": 141510
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.925921976632596e-05,
+      "loss": 0.9132,
+      "step": 141520
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9258458116897955e-05,
+      "loss": 1.0285,
+      "step": 141530
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.925769646746996e-05,
+      "loss": 0.8565,
+      "step": 141540
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.925693481804195e-05,
+      "loss": 0.9359,
+      "step": 141550
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9256173168613954e-05,
+      "loss": 0.9794,
+      "step": 141560
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.925541151918595e-05,
+      "loss": 0.9156,
+      "step": 141570
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.925464986975795e-05,
+      "loss": 0.8757,
+      "step": 141580
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.925388822032995e-05,
+      "loss": 0.9196,
+      "step": 141590
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.925312657090195e-05,
+      "loss": 0.8414,
+      "step": 141600
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.925236492147394e-05,
+      "loss": 0.9161,
+      "step": 141610
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9251603272045946e-05,
+      "loss": 0.8687,
+      "step": 141620
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9250841622617943e-05,
+      "loss": 0.9122,
+      "step": 141630
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.925007997318994e-05,
+      "loss": 0.8624,
+      "step": 141640
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924931832376194e-05,
+      "loss": 0.9084,
+      "step": 141650
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924855667433394e-05,
+      "loss": 0.9474,
+      "step": 141660
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924779502490593e-05,
+      "loss": 0.9712,
+      "step": 141670
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924703337547794e-05,
+      "loss": 0.874,
+      "step": 141680
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9246271726049935e-05,
+      "loss": 1.0122,
+      "step": 141690
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924551007662193e-05,
+      "loss": 0.9341,
+      "step": 141700
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924474842719394e-05,
+      "loss": 0.9049,
+      "step": 141710
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9243986777765934e-05,
+      "loss": 1.0465,
+      "step": 141720
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924322512833793e-05,
+      "loss": 0.9924,
+      "step": 141730
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924246347890993e-05,
+      "loss": 0.9952,
+      "step": 141740
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9241701829481933e-05,
+      "loss": 0.8553,
+      "step": 141750
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9240940180053924e-05,
+      "loss": 1.0639,
+      "step": 141760
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.924017853062593e-05,
+      "loss": 1.0025,
+      "step": 141770
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9239416881197926e-05,
+      "loss": 0.8948,
+      "step": 141780
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.923865523176992e-05,
+      "loss": 0.9454,
+      "step": 141790
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.923789358234192e-05,
+      "loss": 1.0019,
+      "step": 141800
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923713193291392e-05,
+      "loss": 0.9458,
+      "step": 141810
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.9236370283485916e-05,
+      "loss": 0.9587,
+      "step": 141820
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923560863405792e-05,
+      "loss": 0.8916,
+      "step": 141830
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923484698462991e-05,
+      "loss": 0.9647,
+      "step": 141840
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.9234085335201915e-05,
+      "loss": 0.9665,
+      "step": 141850
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923332368577391e-05,
+      "loss": 1.0088,
+      "step": 141860
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923256203634591e-05,
+      "loss": 0.9355,
+      "step": 141870
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.9231800386917914e-05,
+      "loss": 0.8979,
+      "step": 141880
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923103873748991e-05,
+      "loss": 0.9975,
+      "step": 141890
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.923027708806191e-05,
+      "loss": 0.9763,
+      "step": 141900
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.9229515438633906e-05,
+      "loss": 0.9727,
+      "step": 141910
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.922875378920591e-05,
+      "loss": 0.9929,
+      "step": 141920
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.92279921397779e-05,
+      "loss": 0.9964,
+      "step": 141930
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.9227230490349906e-05,
+      "loss": 0.8545,
+      "step": 141940
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.92264688409219e-05,
+      "loss": 0.9372,
+      "step": 141950
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.92257071914939e-05,
+      "loss": 0.9648,
+      "step": 141960
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.92249455420659e-05,
+      "loss": 0.94,
+      "step": 141970
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.92241838926379e-05,
+      "loss": 1.018,
+      "step": 141980
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.922342224320989e-05,
+      "loss": 0.9165,
+      "step": 141990
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.92226605937819e-05,
+      "loss": 0.8924,
+      "step": 142000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.9221898944353895e-05,
+      "loss": 0.8954,
+      "step": 142010
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.922113729492589e-05,
+      "loss": 0.9272,
+      "step": 142020
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.922037564549789e-05,
+      "loss": 0.9305,
+      "step": 142030
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9219613996069894e-05,
+      "loss": 1.1283,
+      "step": 142040
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9218852346641885e-05,
+      "loss": 0.9477,
+      "step": 142050
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.921809069721389e-05,
+      "loss": 0.9311,
+      "step": 142060
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9217329047785886e-05,
+      "loss": 0.9476,
+      "step": 142070
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9216567398357884e-05,
+      "loss": 0.8962,
+      "step": 142080
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.921580574892989e-05,
+      "loss": 0.91,
+      "step": 142090
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9215044099501885e-05,
+      "loss": 0.965,
+      "step": 142100
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.921428245007388e-05,
+      "loss": 0.9547,
+      "step": 142110
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.921352080064588e-05,
+      "loss": 0.8795,
+      "step": 142120
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9212759151217885e-05,
+      "loss": 0.9694,
+      "step": 142130
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9211997501789875e-05,
+      "loss": 0.8774,
+      "step": 142140
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.921123585236188e-05,
+      "loss": 0.8832,
+      "step": 142150
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.921047420293388e-05,
+      "loss": 0.9619,
+      "step": 142160
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9209712553505874e-05,
+      "loss": 0.9218,
+      "step": 142170
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.920895090407787e-05,
+      "loss": 0.911,
+      "step": 142180
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9208189254649876e-05,
+      "loss": 0.9219,
+      "step": 142190
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.920742760522187e-05,
+      "loss": 0.8624,
+      "step": 142200
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.920666595579387e-05,
+      "loss": 0.9554,
+      "step": 142210
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.920590430636587e-05,
+      "loss": 0.8283,
+      "step": 142220
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9205142656937866e-05,
+      "loss": 1.0693,
+      "step": 142230
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9204381007509864e-05,
+      "loss": 0.9531,
+      "step": 142240
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.920361935808187e-05,
+      "loss": 0.9146,
+      "step": 142250
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.920285770865386e-05,
+      "loss": 0.8603,
+      "step": 142260
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.920209605922586e-05,
+      "loss": 0.9023,
+      "step": 142270
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.920133440979786e-05,
+      "loss": 0.862,
+      "step": 142280
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.920057276036986e-05,
+      "loss": 0.8745,
+      "step": 142290
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919981111094186e-05,
+      "loss": 0.9346,
+      "step": 142300
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919904946151385e-05,
+      "loss": 0.9351,
+      "step": 142310
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919828781208586e-05,
+      "loss": 1.021,
+      "step": 142320
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9197526162657854e-05,
+      "loss": 0.8676,
+      "step": 142330
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919676451322985e-05,
+      "loss": 0.8797,
+      "step": 142340
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919600286380185e-05,
+      "loss": 0.9049,
+      "step": 142350
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9195241214373853e-05,
+      "loss": 0.9382,
+      "step": 142360
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9194479564945844e-05,
+      "loss": 0.8812,
+      "step": 142370
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919371791551785e-05,
+      "loss": 0.9288,
+      "step": 142380
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9192956266089846e-05,
+      "loss": 0.9531,
+      "step": 142390
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919219461666184e-05,
+      "loss": 0.998,
+      "step": 142400
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.919143296723384e-05,
+      "loss": 1.0096,
+      "step": 142410
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9190671317805845e-05,
+      "loss": 1.0849,
+      "step": 142420
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9189909668377836e-05,
+      "loss": 0.9298,
+      "step": 142430
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.918914801894984e-05,
+      "loss": 0.8661,
+      "step": 142440
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.918838636952184e-05,
+      "loss": 1.0061,
+      "step": 142450
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9187624720093835e-05,
+      "loss": 0.9098,
+      "step": 142460
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.918686307066583e-05,
+      "loss": 1.0067,
+      "step": 142470
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9186101421237837e-05,
+      "loss": 0.8638,
+      "step": 142480
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9185339771809834e-05,
+      "loss": 0.9317,
+      "step": 142490
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.918457812238183e-05,
+      "loss": 0.8816,
+      "step": 142500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9183816472953836e-05,
+      "loss": 0.9566,
+      "step": 142510
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9183054823525826e-05,
+      "loss": 0.8637,
+      "step": 142520
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.918229317409783e-05,
+      "loss": 0.8921,
+      "step": 142530
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.918153152466983e-05,
+      "loss": 0.9455,
+      "step": 142540
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9180769875241826e-05,
+      "loss": 0.9206,
+      "step": 142550
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.918000822581382e-05,
+      "loss": 0.8304,
+      "step": 142560
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917924657638583e-05,
+      "loss": 0.9636,
+      "step": 142570
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917848492695782e-05,
+      "loss": 0.8719,
+      "step": 142580
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917772327752982e-05,
+      "loss": 0.8727,
+      "step": 142590
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917696162810182e-05,
+      "loss": 0.9687,
+      "step": 142600
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917619997867382e-05,
+      "loss": 0.8671,
+      "step": 142610
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9175438329245815e-05,
+      "loss": 0.911,
+      "step": 142620
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917467667981782e-05,
+      "loss": 0.9332,
+      "step": 142630
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917391503038981e-05,
+      "loss": 0.9577,
+      "step": 142640
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9173153380961814e-05,
+      "loss": 0.9401,
+      "step": 142650
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917239173153381e-05,
+      "loss": 0.9435,
+      "step": 142660
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.917163008210581e-05,
+      "loss": 0.8033,
+      "step": 142670
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.917086843267781e-05,
+      "loss": 0.8464,
+      "step": 142680
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.917010678324981e-05,
+      "loss": 0.9863,
+      "step": 142690
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.916934513382181e-05,
+      "loss": 0.7922,
+      "step": 142700
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9168583484393805e-05,
+      "loss": 0.9731,
+      "step": 142710
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.91678218349658e-05,
+      "loss": 0.8153,
+      "step": 142720
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.91670601855378e-05,
+      "loss": 0.988,
+      "step": 142730
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9166298536109805e-05,
+      "loss": 0.9699,
+      "step": 142740
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9165536886681795e-05,
+      "loss": 0.8523,
+      "step": 142750
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.91647752372538e-05,
+      "loss": 0.9602,
+      "step": 142760
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.91640135878258e-05,
+      "loss": 0.978,
+      "step": 142770
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9163251938397794e-05,
+      "loss": 0.8772,
+      "step": 142780
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.916249028896979e-05,
+      "loss": 0.9259,
+      "step": 142790
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9161728639541796e-05,
+      "loss": 0.9353,
+      "step": 142800
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.916096699011379e-05,
+      "loss": 0.9126,
+      "step": 142810
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.916020534068579e-05,
+      "loss": 0.7699,
+      "step": 142820
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.915944369125779e-05,
+      "loss": 0.9793,
+      "step": 142830
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9158682041829786e-05,
+      "loss": 0.92,
+      "step": 142840
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9157920392401784e-05,
+      "loss": 0.9232,
+      "step": 142850
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.915715874297379e-05,
+      "loss": 0.8483,
+      "step": 142860
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9156397093545785e-05,
+      "loss": 0.8108,
+      "step": 142870
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.915563544411778e-05,
+      "loss": 0.9399,
+      "step": 142880
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.915487379468979e-05,
+      "loss": 0.9385,
+      "step": 142890
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.915411214526178e-05,
+      "loss": 0.8678,
+      "step": 142900
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.915335049583378e-05,
+      "loss": 1.0187,
+      "step": 142910
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.915258884640578e-05,
+      "loss": 0.8656,
+      "step": 142920
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.915182719697778e-05,
+      "loss": 1.0404,
+      "step": 142930
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.9151065547549774e-05,
+      "loss": 1.0816,
+      "step": 142940
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.915030389812178e-05,
+      "loss": 0.8294,
+      "step": 142950
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914954224869377e-05,
+      "loss": 0.92,
+      "step": 142960
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.9148780599265773e-05,
+      "loss": 1.1394,
+      "step": 142970
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914801894983777e-05,
+      "loss": 0.9726,
+      "step": 142980
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914725730040977e-05,
+      "loss": 1.0153,
+      "step": 142990
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.9146495650981766e-05,
+      "loss": 0.926,
+      "step": 143000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914573400155377e-05,
+      "loss": 0.9422,
+      "step": 143010
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914497235212576e-05,
+      "loss": 0.9172,
+      "step": 143020
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.9144210702697765e-05,
+      "loss": 0.8892,
+      "step": 143030
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914344905326976e-05,
+      "loss": 0.8962,
+      "step": 143040
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914268740384176e-05,
+      "loss": 0.9753,
+      "step": 143050
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914192575441376e-05,
+      "loss": 0.9677,
+      "step": 143060
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914116410498576e-05,
+      "loss": 0.9956,
+      "step": 143070
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.914040245555776e-05,
+      "loss": 0.9238,
+      "step": 143080
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.9139640806129757e-05,
+      "loss": 0.9064,
+      "step": 143090
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.913887915670176e-05,
+      "loss": 0.9239,
+      "step": 143100
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.913811750727375e-05,
+      "loss": 0.8591,
+      "step": 143110
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9137355857845756e-05,
+      "loss": 0.9507,
+      "step": 143120
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913659420841775e-05,
+      "loss": 0.8531,
+      "step": 143130
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913583255898975e-05,
+      "loss": 0.9795,
+      "step": 143140
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913507090956175e-05,
+      "loss": 1.0634,
+      "step": 143150
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913430926013375e-05,
+      "loss": 0.9055,
+      "step": 143160
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913354761070574e-05,
+      "loss": 0.9952,
+      "step": 143170
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913278596127775e-05,
+      "loss": 0.9957,
+      "step": 143180
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913202431184974e-05,
+      "loss": 1.0003,
+      "step": 143190
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913126266242174e-05,
+      "loss": 0.9068,
+      "step": 143200
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.913050101299374e-05,
+      "loss": 0.9129,
+      "step": 143210
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912973936356574e-05,
+      "loss": 0.9624,
+      "step": 143220
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9128977714137735e-05,
+      "loss": 0.8854,
+      "step": 143230
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912821606470974e-05,
+      "loss": 0.8852,
+      "step": 143240
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9127454415281736e-05,
+      "loss": 0.9217,
+      "step": 143250
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9126692765853734e-05,
+      "loss": 0.9509,
+      "step": 143260
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912593111642574e-05,
+      "loss": 0.8935,
+      "step": 143270
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912516946699773e-05,
+      "loss": 0.8771,
+      "step": 143280
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912440781756973e-05,
+      "loss": 0.9473,
+      "step": 143290
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912364616814173e-05,
+      "loss": 1.0133,
+      "step": 143300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912288451871373e-05,
+      "loss": 0.9361,
+      "step": 143310
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9122122869285725e-05,
+      "loss": 0.8958,
+      "step": 143320
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.912136121985773e-05,
+      "loss": 0.9795,
+      "step": 143330
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.912059957042972e-05,
+      "loss": 0.8455,
+      "step": 143340
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9119837921001725e-05,
+      "loss": 0.8808,
+      "step": 143350
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911907627157372e-05,
+      "loss": 0.899,
+      "step": 143360
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911831462214572e-05,
+      "loss": 0.9875,
+      "step": 143370
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911755297271772e-05,
+      "loss": 0.8847,
+      "step": 143380
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911679132328972e-05,
+      "loss": 0.9806,
+      "step": 143390
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911602967386171e-05,
+      "loss": 0.9721,
+      "step": 143400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9115268024433716e-05,
+      "loss": 0.9671,
+      "step": 143410
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9114506375005714e-05,
+      "loss": 0.9954,
+      "step": 143420
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911374472557771e-05,
+      "loss": 0.9674,
+      "step": 143430
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911298307614971e-05,
+      "loss": 0.9457,
+      "step": 143440
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911222142672171e-05,
+      "loss": 1.007,
+      "step": 143450
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911145977729371e-05,
+      "loss": 0.9337,
+      "step": 143460
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.911069812786571e-05,
+      "loss": 0.9719,
+      "step": 143470
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.910993647843771e-05,
+      "loss": 0.9701,
+      "step": 143480
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.91091748290097e-05,
+      "loss": 0.8559,
+      "step": 143490
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.910841317958171e-05,
+      "loss": 0.9469,
+      "step": 143500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9107651530153704e-05,
+      "loss": 1.0024,
+      "step": 143510
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.91068898807257e-05,
+      "loss": 0.8518,
+      "step": 143520
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.91061282312977e-05,
+      "loss": 0.9196,
+      "step": 143530
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9105366581869704e-05,
+      "loss": 0.881,
+      "step": 143540
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9104604932441694e-05,
+      "loss": 0.8989,
+      "step": 143550
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.91038432830137e-05,
+      "loss": 0.911,
+      "step": 143560
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9103081633585696e-05,
+      "loss": 0.9957,
+      "step": 143570
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9102319984157693e-05,
+      "loss": 0.785,
+      "step": 143580
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.910155833472969e-05,
+      "loss": 0.8436,
+      "step": 143590
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9100796685301695e-05,
+      "loss": 0.9148,
+      "step": 143600
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9100035035873686e-05,
+      "loss": 1.0827,
+      "step": 143610
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909927338644569e-05,
+      "loss": 1.0282,
+      "step": 143620
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909851173701769e-05,
+      "loss": 0.9891,
+      "step": 143630
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9097750087589685e-05,
+      "loss": 0.9822,
+      "step": 143640
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909698843816168e-05,
+      "loss": 1.0079,
+      "step": 143650
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909622678873368e-05,
+      "loss": 1.095,
+      "step": 143660
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9095465139305684e-05,
+      "loss": 0.9045,
+      "step": 143670
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909470348987768e-05,
+      "loss": 0.9599,
+      "step": 143680
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909394184044968e-05,
+      "loss": 0.9362,
+      "step": 143690
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9093180191021677e-05,
+      "loss": 1.0103,
+      "step": 143700
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909241854159368e-05,
+      "loss": 0.9975,
+      "step": 143710
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909165689216567e-05,
+      "loss": 0.892,
+      "step": 143720
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9090895242737676e-05,
+      "loss": 0.9828,
+      "step": 143730
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.909013359330967e-05,
+      "loss": 0.8761,
+      "step": 143740
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.908937194388167e-05,
+      "loss": 0.8887,
+      "step": 143750
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.908861029445367e-05,
+      "loss": 0.8665,
+      "step": 143760
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.908784864502567e-05,
+      "loss": 0.9951,
+      "step": 143770
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.908708699559766e-05,
+      "loss": 0.8581,
+      "step": 143780
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.908632534616967e-05,
+      "loss": 0.9005,
+      "step": 143790
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9085563696741665e-05,
+      "loss": 0.9478,
+      "step": 143800
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.908480204731366e-05,
+      "loss": 0.9229,
+      "step": 143810
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.908404039788566e-05,
+      "loss": 0.9102,
+      "step": 143820
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9083278748457664e-05,
+      "loss": 0.8866,
+      "step": 143830
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.908251709902966e-05,
+      "loss": 0.8647,
+      "step": 143840
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.908175544960166e-05,
+      "loss": 1.0051,
+      "step": 143850
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9080993800173656e-05,
+      "loss": 1.0704,
+      "step": 143860
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9080232150745654e-05,
+      "loss": 0.9746,
+      "step": 143870
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907947050131766e-05,
+      "loss": 0.9157,
+      "step": 143880
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9078708851889656e-05,
+      "loss": 0.9171,
+      "step": 143890
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907794720246165e-05,
+      "loss": 0.9086,
+      "step": 143900
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907718555303365e-05,
+      "loss": 0.9763,
+      "step": 143910
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9076423903605655e-05,
+      "loss": 0.8892,
+      "step": 143920
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9075662254177645e-05,
+      "loss": 0.9036,
+      "step": 143930
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907490060474965e-05,
+      "loss": 1.0195,
+      "step": 143940
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907413895532165e-05,
+      "loss": 0.9128,
+      "step": 143950
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9073377305893645e-05,
+      "loss": 0.9392,
+      "step": 143960
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907261565646564e-05,
+      "loss": 1.006,
+      "step": 143970
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9071854007037646e-05,
+      "loss": 0.9705,
+      "step": 143980
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.907109235760964e-05,
+      "loss": 1.0395,
+      "step": 143990
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.907033070818164e-05,
+      "loss": 0.8443,
+      "step": 144000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906956905875364e-05,
+      "loss": 0.9938,
+      "step": 144010
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9068807409325636e-05,
+      "loss": 1.0102,
+      "step": 144020
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9068045759897634e-05,
+      "loss": 1.0884,
+      "step": 144030
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906728411046964e-05,
+      "loss": 0.9174,
+      "step": 144040
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9066522461041635e-05,
+      "loss": 0.9368,
+      "step": 144050
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906576081161363e-05,
+      "loss": 0.8953,
+      "step": 144060
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906499916218564e-05,
+      "loss": 0.9559,
+      "step": 144070
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906423751275763e-05,
+      "loss": 0.8909,
+      "step": 144080
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906347586332963e-05,
+      "loss": 0.8485,
+      "step": 144090
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906271421390163e-05,
+      "loss": 0.8828,
+      "step": 144100
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906195256447363e-05,
+      "loss": 1.0173,
+      "step": 144110
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9061190915045624e-05,
+      "loss": 0.9222,
+      "step": 144120
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.906042926561762e-05,
+      "loss": 0.9965,
+      "step": 144130
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.905966761618962e-05,
+      "loss": 0.7778,
+      "step": 144140
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9058905966761624e-05,
+      "loss": 0.9445,
+      "step": 144150
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9058144317333614e-05,
+      "loss": 0.958,
+      "step": 144160
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.905738266790562e-05,
+      "loss": 0.8735,
+      "step": 144170
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9056621018477616e-05,
+      "loss": 0.9358,
+      "step": 144180
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.9055859369049613e-05,
+      "loss": 0.9021,
+      "step": 144190
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.905509771962161e-05,
+      "loss": 0.8711,
+      "step": 144200
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9054336070193615e-05,
+      "loss": 1.0176,
+      "step": 144210
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.905357442076561e-05,
+      "loss": 0.9538,
+      "step": 144220
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.905281277133761e-05,
+      "loss": 0.975,
+      "step": 144230
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.905205112190961e-05,
+      "loss": 0.9225,
+      "step": 144240
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9051289472481605e-05,
+      "loss": 0.9591,
+      "step": 144250
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.905052782305361e-05,
+      "loss": 0.9601,
+      "step": 144260
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.904976617362561e-05,
+      "loss": 0.9742,
+      "step": 144270
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9049004524197604e-05,
+      "loss": 0.8869,
+      "step": 144280
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.90482428747696e-05,
+      "loss": 0.9447,
+      "step": 144290
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9047481225341606e-05,
+      "loss": 0.9636,
+      "step": 144300
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.90467195759136e-05,
+      "loss": 1.0156,
+      "step": 144310
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.90459579264856e-05,
+      "loss": 0.9704,
+      "step": 144320
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.90451962770576e-05,
+      "loss": 0.9032,
+      "step": 144330
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9044434627629596e-05,
+      "loss": 0.9595,
+      "step": 144340
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.904367297820159e-05,
+      "loss": 0.9782,
+      "step": 144350
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.90429113287736e-05,
+      "loss": 0.9704,
+      "step": 144360
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.904214967934559e-05,
+      "loss": 0.9114,
+      "step": 144370
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.904138802991759e-05,
+      "loss": 1.0091,
+      "step": 144380
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.904062638048959e-05,
+      "loss": 0.9546,
+      "step": 144390
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.903986473106159e-05,
+      "loss": 0.8994,
+      "step": 144400
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9039103081633585e-05,
+      "loss": 0.931,
+      "step": 144410
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.903834143220559e-05,
+      "loss": 0.8846,
+      "step": 144420
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9037579782777587e-05,
+      "loss": 1.0033,
+      "step": 144430
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9036818133349584e-05,
+      "loss": 0.9106,
+      "step": 144440
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903605648392158e-05,
+      "loss": 0.8467,
+      "step": 144450
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903529483449358e-05,
+      "loss": 0.9041,
+      "step": 144460
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903453318506558e-05,
+      "loss": 0.9527,
+      "step": 144470
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903377153563758e-05,
+      "loss": 0.9613,
+      "step": 144480
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903300988620958e-05,
+      "loss": 0.9731,
+      "step": 144490
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9032248236781576e-05,
+      "loss": 0.8765,
+      "step": 144500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903148658735358e-05,
+      "loss": 0.8808,
+      "step": 144510
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.903072493792557e-05,
+      "loss": 0.9377,
+      "step": 144520
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9029963288497575e-05,
+      "loss": 0.884,
+      "step": 144530
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902920163906957e-05,
+      "loss": 0.9062,
+      "step": 144540
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902843998964157e-05,
+      "loss": 0.9448,
+      "step": 144550
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902767834021357e-05,
+      "loss": 0.9737,
+      "step": 144560
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9026916690785565e-05,
+      "loss": 0.8709,
+      "step": 144570
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902615504135756e-05,
+      "loss": 0.8914,
+      "step": 144580
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9025393391929566e-05,
+      "loss": 0.9736,
+      "step": 144590
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9024631742501564e-05,
+      "loss": 0.9252,
+      "step": 144600
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902387009307356e-05,
+      "loss": 0.9434,
+      "step": 144610
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902310844364556e-05,
+      "loss": 1.0039,
+      "step": 144620
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.9022346794217556e-05,
+      "loss": 1.025,
+      "step": 144630
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.902158514478956e-05,
+      "loss": 0.8592,
+      "step": 144640
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.902082349536156e-05,
+      "loss": 0.9459,
+      "step": 144650
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9020061845933555e-05,
+      "loss": 0.9219,
+      "step": 144660
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901930019650555e-05,
+      "loss": 0.8993,
+      "step": 144670
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901853854707756e-05,
+      "loss": 0.9665,
+      "step": 144680
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901777689764955e-05,
+      "loss": 0.9348,
+      "step": 144690
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901701524822155e-05,
+      "loss": 0.972,
+      "step": 144700
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901625359879355e-05,
+      "loss": 0.9444,
+      "step": 144710
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901549194936555e-05,
+      "loss": 0.9987,
+      "step": 144720
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9014730299937544e-05,
+      "loss": 0.9251,
+      "step": 144730
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901396865050955e-05,
+      "loss": 0.9778,
+      "step": 144740
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901320700108154e-05,
+      "loss": 1.0263,
+      "step": 144750
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9012445351653544e-05,
+      "loss": 0.9744,
+      "step": 144760
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901168370222554e-05,
+      "loss": 0.8617,
+      "step": 144770
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.901092205279754e-05,
+      "loss": 0.9057,
+      "step": 144780
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9010160403369536e-05,
+      "loss": 0.9373,
+      "step": 144790
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.900939875394154e-05,
+      "loss": 0.821,
+      "step": 144800
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.900863710451354e-05,
+      "loss": 0.9304,
+      "step": 144810
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9007875455085535e-05,
+      "loss": 0.9472,
+      "step": 144820
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.900711380565753e-05,
+      "loss": 1.0545,
+      "step": 144830
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.900635215622953e-05,
+      "loss": 0.951,
+      "step": 144840
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9005590506801534e-05,
+      "loss": 1.0401,
+      "step": 144850
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.900482885737353e-05,
+      "loss": 0.9538,
+      "step": 144860
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.900406720794553e-05,
+      "loss": 0.9698,
+      "step": 144870
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.900330555851753e-05,
+      "loss": 0.9784,
+      "step": 144880
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.900254390908953e-05,
+      "loss": 1.0945,
+      "step": 144890
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.900178225966152e-05,
+      "loss": 0.9026,
+      "step": 144900
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.9001020610233526e-05,
+      "loss": 0.942,
+      "step": 144910
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.9000258960805523e-05,
+      "loss": 0.8687,
+      "step": 144920
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899949731137752e-05,
+      "loss": 0.9368,
+      "step": 144930
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899873566194952e-05,
+      "loss": 0.8478,
+      "step": 144940
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899797401252152e-05,
+      "loss": 0.9037,
+      "step": 144950
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899721236309351e-05,
+      "loss": 0.9462,
+      "step": 144960
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899645071366552e-05,
+      "loss": 0.9397,
+      "step": 144970
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8995689064237515e-05,
+      "loss": 0.9158,
+      "step": 144980
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899492741480951e-05,
+      "loss": 0.9861,
+      "step": 144990
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899416576538151e-05,
+      "loss": 1.0578,
+      "step": 145000
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8993404115953514e-05,
+      "loss": 0.8742,
+      "step": 145010
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899264246652551e-05,
+      "loss": 0.8686,
+      "step": 145020
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.899188081709751e-05,
+      "loss": 1.0011,
+      "step": 145030
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8991119167669507e-05,
+      "loss": 0.8842,
+      "step": 145040
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8990357518241504e-05,
+      "loss": 0.9819,
+      "step": 145050
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.898959586881351e-05,
+      "loss": 0.9948,
+      "step": 145060
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.89888342193855e-05,
+      "loss": 0.8949,
+      "step": 145070
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.89880725699575e-05,
+      "loss": 1.0315,
+      "step": 145080
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.89873109205295e-05,
+      "loss": 1.0177,
+      "step": 145090
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.89865492711015e-05,
+      "loss": 0.8434,
+      "step": 145100
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8985787621673496e-05,
+      "loss": 0.8817,
+      "step": 145110
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.89850259722455e-05,
+      "loss": 0.9512,
+      "step": 145120
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.898426432281749e-05,
+      "loss": 0.9296,
+      "step": 145130
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8983502673389495e-05,
+      "loss": 0.7896,
+      "step": 145140
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.898274102396149e-05,
+      "loss": 0.9082,
+      "step": 145150
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.898197937453349e-05,
+      "loss": 0.9322,
+      "step": 145160
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.898121772510549e-05,
+      "loss": 0.9796,
+      "step": 145170
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.898045607567749e-05,
+      "loss": 0.8935,
+      "step": 145180
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897969442624949e-05,
+      "loss": 0.9769,
+      "step": 145190
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8978932776821486e-05,
+      "loss": 0.7669,
+      "step": 145200
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8978171127393484e-05,
+      "loss": 0.926,
+      "step": 145210
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897740947796548e-05,
+      "loss": 0.9406,
+      "step": 145220
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8976647828537486e-05,
+      "loss": 0.9389,
+      "step": 145230
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897588617910948e-05,
+      "loss": 0.9265,
+      "step": 145240
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897512452968148e-05,
+      "loss": 0.9596,
+      "step": 145250
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897436288025348e-05,
+      "loss": 0.9642,
+      "step": 145260
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897360123082548e-05,
+      "loss": 0.917,
+      "step": 145270
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897283958139747e-05,
+      "loss": 0.9344,
+      "step": 145280
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.897207793196948e-05,
+      "loss": 0.974,
+      "step": 145290
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8971316282541475e-05,
+      "loss": 0.9742,
+      "step": 145300
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.897055463311347e-05,
+      "loss": 0.8638,
+      "step": 145310
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896979298368547e-05,
+      "loss": 0.9193,
+      "step": 145320
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8969031334257474e-05,
+      "loss": 0.9879,
+      "step": 145330
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8968269684829464e-05,
+      "loss": 0.9019,
+      "step": 145340
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896750803540147e-05,
+      "loss": 0.9197,
+      "step": 145350
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8966746385973466e-05,
+      "loss": 0.983,
+      "step": 145360
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8965984736545464e-05,
+      "loss": 0.8777,
+      "step": 145370
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896522308711746e-05,
+      "loss": 0.8829,
+      "step": 145380
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8964461437689465e-05,
+      "loss": 0.9007,
+      "step": 145390
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896369978826146e-05,
+      "loss": 1.0024,
+      "step": 145400
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896293813883346e-05,
+      "loss": 0.9363,
+      "step": 145410
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896217648940546e-05,
+      "loss": 0.8692,
+      "step": 145420
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8961414839977455e-05,
+      "loss": 0.9668,
+      "step": 145430
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.896065319054946e-05,
+      "loss": 0.8986,
+      "step": 145440
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.895989154112146e-05,
+      "loss": 0.9581,
+      "step": 145450
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8959129891693454e-05,
+      "loss": 0.9722,
+      "step": 145460
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.895836824226545e-05,
+      "loss": 0.9576,
+      "step": 145470
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8957606592837456e-05,
+      "loss": 0.99,
+      "step": 145480
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.895684494340945e-05,
+      "loss": 0.9517,
+      "step": 145490
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.895608329398145e-05,
+      "loss": 0.903,
+      "step": 145500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.895532164455344e-05,
+      "loss": 0.9947,
+      "step": 145510
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.8954559995125446e-05,
+      "loss": 0.9664,
+      "step": 145520
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8953798345697443e-05,
+      "loss": 0.8697,
+      "step": 145530
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.895303669626944e-05,
+      "loss": 1.0032,
+      "step": 145540
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.895227504684144e-05,
+      "loss": 0.8864,
+      "step": 145550
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.895151339741344e-05,
+      "loss": 0.9009,
+      "step": 145560
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.895075174798544e-05,
+      "loss": 0.8927,
+      "step": 145570
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894999009855744e-05,
+      "loss": 0.9224,
+      "step": 145580
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8949228449129435e-05,
+      "loss": 0.8788,
+      "step": 145590
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894846679970143e-05,
+      "loss": 0.8968,
+      "step": 145600
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894770515027344e-05,
+      "loss": 0.8662,
+      "step": 145610
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8946943500845434e-05,
+      "loss": 1.0086,
+      "step": 145620
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894618185141743e-05,
+      "loss": 0.9425,
+      "step": 145630
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894542020198943e-05,
+      "loss": 0.9873,
+      "step": 145640
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894465855256143e-05,
+      "loss": 1.0013,
+      "step": 145650
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8943896903133424e-05,
+      "loss": 1.0514,
+      "step": 145660
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894313525370543e-05,
+      "loss": 0.9441,
+      "step": 145670
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8942373604277426e-05,
+      "loss": 0.9593,
+      "step": 145680
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894161195484942e-05,
+      "loss": 0.8873,
+      "step": 145690
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.894085030542142e-05,
+      "loss": 0.9726,
+      "step": 145700
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8940088655993425e-05,
+      "loss": 0.9791,
+      "step": 145710
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8939327006565416e-05,
+      "loss": 0.848,
+      "step": 145720
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.893856535713742e-05,
+      "loss": 0.9543,
+      "step": 145730
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.893780370770942e-05,
+      "loss": 0.8842,
+      "step": 145740
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8937042058281415e-05,
+      "loss": 1.0271,
+      "step": 145750
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.893628040885341e-05,
+      "loss": 0.9756,
+      "step": 145760
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8935518759425417e-05,
+      "loss": 0.8508,
+      "step": 145770
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8934757109997414e-05,
+      "loss": 0.8219,
+      "step": 145780
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.893399546056941e-05,
+      "loss": 0.9833,
+      "step": 145790
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.893323381114141e-05,
+      "loss": 0.9522,
+      "step": 145800
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8932472161713406e-05,
+      "loss": 0.9741,
+      "step": 145810
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.893171051228541e-05,
+      "loss": 0.8709,
+      "step": 145820
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.893094886285741e-05,
+      "loss": 0.873,
+      "step": 145830
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8930187213429406e-05,
+      "loss": 0.8784,
+      "step": 145840
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.89294255640014e-05,
+      "loss": 0.9516,
+      "step": 145850
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.892866391457341e-05,
+      "loss": 0.9122,
+      "step": 145860
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.89279022651454e-05,
+      "loss": 0.9632,
+      "step": 145870
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.89271406157174e-05,
+      "loss": 1.0154,
+      "step": 145880
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.89263789662894e-05,
+      "loss": 0.8958,
+      "step": 145890
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.89256173168614e-05,
+      "loss": 0.8614,
+      "step": 145900
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8924855667433395e-05,
+      "loss": 1.0297,
+      "step": 145910
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.89240940180054e-05,
+      "loss": 1.0015,
+      "step": 145920
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.892333236857739e-05,
+      "loss": 0.8928,
+      "step": 145930
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8922570719149394e-05,
+      "loss": 0.8735,
+      "step": 145940
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.8921809069721384e-05,
+      "loss": 1.014,
+      "step": 145950
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.892104742029339e-05,
+      "loss": 0.9841,
+      "step": 145960
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8920285770865386e-05,
+      "loss": 0.9733,
+      "step": 145970
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8919524121437384e-05,
+      "loss": 1.0673,
+      "step": 145980
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891876247200939e-05,
+      "loss": 0.932,
+      "step": 145990
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8918000822581385e-05,
+      "loss": 0.9301,
+      "step": 146000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891723917315338e-05,
+      "loss": 0.9838,
+      "step": 146010
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891647752372538e-05,
+      "loss": 0.918,
+      "step": 146020
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8915715874297385e-05,
+      "loss": 0.9349,
+      "step": 146030
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8914954224869375e-05,
+      "loss": 0.8393,
+      "step": 146040
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891419257544138e-05,
+      "loss": 0.9003,
+      "step": 146050
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891343092601338e-05,
+      "loss": 0.9756,
+      "step": 146060
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8912669276585374e-05,
+      "loss": 1.0109,
+      "step": 146070
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891190762715737e-05,
+      "loss": 0.9844,
+      "step": 146080
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8911145977729376e-05,
+      "loss": 0.8964,
+      "step": 146090
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891038432830137e-05,
+      "loss": 0.9653,
+      "step": 146100
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.890962267887337e-05,
+      "loss": 0.8944,
+      "step": 146110
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.890886102944537e-05,
+      "loss": 0.8363,
+      "step": 146120
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8908099380017366e-05,
+      "loss": 0.9996,
+      "step": 146130
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8907337730589363e-05,
+      "loss": 0.9134,
+      "step": 146140
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.890657608116137e-05,
+      "loss": 0.8576,
+      "step": 146150
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.890581443173336e-05,
+      "loss": 1.009,
+      "step": 146160
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.890505278230536e-05,
+      "loss": 0.9492,
+      "step": 146170
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.890429113287736e-05,
+      "loss": 0.9043,
+      "step": 146180
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.890352948344936e-05,
+      "loss": 0.8431,
+      "step": 146190
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.890276783402136e-05,
+      "loss": 1.0193,
+      "step": 146200
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.890200618459336e-05,
+      "loss": 0.9346,
+      "step": 146210
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.890124453516536e-05,
+      "loss": 0.8645,
+      "step": 146220
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.8900482885737354e-05,
+      "loss": 1.0128,
+      "step": 146230
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889972123630936e-05,
+      "loss": 1.0544,
+      "step": 146240
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889895958688135e-05,
+      "loss": 0.92,
+      "step": 146250
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889819793745335e-05,
+      "loss": 1.0733,
+      "step": 146260
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889743628802535e-05,
+      "loss": 0.9561,
+      "step": 146270
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889667463859735e-05,
+      "loss": 0.9488,
+      "step": 146280
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.8895912989169346e-05,
+      "loss": 0.9527,
+      "step": 146290
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889515133974135e-05,
+      "loss": 0.9458,
+      "step": 146300
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889438969031334e-05,
+      "loss": 0.8851,
+      "step": 146310
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.8893628040885345e-05,
+      "loss": 0.9919,
+      "step": 146320
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889286639145734e-05,
+      "loss": 0.9795,
+      "step": 146330
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889210474202934e-05,
+      "loss": 0.908,
+      "step": 146340
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889134309260134e-05,
+      "loss": 0.858,
+      "step": 146350
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.889058144317334e-05,
+      "loss": 0.9268,
+      "step": 146360
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.888981979374534e-05,
+      "loss": 0.9741,
+      "step": 146370
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.8889058144317337e-05,
+      "loss": 1.0432,
+      "step": 146380
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.8888296494889334e-05,
+      "loss": 0.902,
+      "step": 146390
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888753484546133e-05,
+      "loss": 1.1468,
+      "step": 146400
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8886773196033336e-05,
+      "loss": 0.8973,
+      "step": 146410
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8886011546605326e-05,
+      "loss": 0.9067,
+      "step": 146420
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888524989717733e-05,
+      "loss": 1.0086,
+      "step": 146430
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888448824774933e-05,
+      "loss": 0.8479,
+      "step": 146440
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8883726598321326e-05,
+      "loss": 1.0128,
+      "step": 146450
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888296494889332e-05,
+      "loss": 0.9305,
+      "step": 146460
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888220329946533e-05,
+      "loss": 0.9622,
+      "step": 146470
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888144165003732e-05,
+      "loss": 0.8805,
+      "step": 146480
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.888068000060932e-05,
+      "loss": 0.9528,
+      "step": 146490
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887991835118132e-05,
+      "loss": 1.0163,
+      "step": 146500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887915670175332e-05,
+      "loss": 0.919,
+      "step": 146510
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8878395052325315e-05,
+      "loss": 1.0268,
+      "step": 146520
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887763340289732e-05,
+      "loss": 0.9622,
+      "step": 146530
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887687175346931e-05,
+      "loss": 0.9801,
+      "step": 146540
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8876110104041314e-05,
+      "loss": 0.9064,
+      "step": 146550
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887534845461331e-05,
+      "loss": 0.987,
+      "step": 146560
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887458680518531e-05,
+      "loss": 1.0217,
+      "step": 146570
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887382515575731e-05,
+      "loss": 0.8696,
+      "step": 146580
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887306350632931e-05,
+      "loss": 0.938,
+      "step": 146590
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.887230185690131e-05,
+      "loss": 0.9947,
+      "step": 146600
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8871540207473305e-05,
+      "loss": 0.8853,
+      "step": 146610
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.887077855804531e-05,
+      "loss": 0.9511,
+      "step": 146620
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.88700169086173e-05,
+      "loss": 1.0004,
+      "step": 146630
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8869255259189305e-05,
+      "loss": 0.933,
+      "step": 146640
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.88684936097613e-05,
+      "loss": 0.9767,
+      "step": 146650
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.88677319603333e-05,
+      "loss": 0.8788,
+      "step": 146660
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.88669703109053e-05,
+      "loss": 1.1131,
+      "step": 146670
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.88662086614773e-05,
+      "loss": 1.0677,
+      "step": 146680
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.886544701204929e-05,
+      "loss": 0.8931,
+      "step": 146690
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8864685362621296e-05,
+      "loss": 0.9509,
+      "step": 146700
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8863923713193294e-05,
+      "loss": 0.928,
+      "step": 146710
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.886316206376529e-05,
+      "loss": 0.9315,
+      "step": 146720
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.886240041433729e-05,
+      "loss": 0.8972,
+      "step": 146730
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.886163876490929e-05,
+      "loss": 0.8727,
+      "step": 146740
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8860877115481283e-05,
+      "loss": 0.9454,
+      "step": 146750
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.886011546605329e-05,
+      "loss": 0.8305,
+      "step": 146760
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8859353816625285e-05,
+      "loss": 0.9612,
+      "step": 146770
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.885859216719728e-05,
+      "loss": 0.9642,
+      "step": 146780
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.885783051776929e-05,
+      "loss": 1.0178,
+      "step": 146790
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8857068868341284e-05,
+      "loss": 0.9125,
+      "step": 146800
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.885630721891328e-05,
+      "loss": 1.0094,
+      "step": 146810
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.885554556948528e-05,
+      "loss": 1.092,
+      "step": 146820
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8854783920057284e-05,
+      "loss": 1.1091,
+      "step": 146830
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8854022270629274e-05,
+      "loss": 0.8544,
+      "step": 146840
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.885326062120128e-05,
+      "loss": 0.9072,
+      "step": 146850
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8852498971773276e-05,
+      "loss": 0.9562,
+      "step": 146860
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.885173732234527e-05,
+      "loss": 0.9369,
+      "step": 146870
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.885097567291727e-05,
+      "loss": 0.9838,
+      "step": 146880
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.885021402348927e-05,
+      "loss": 0.9896,
+      "step": 146890
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8849452374061266e-05,
+      "loss": 0.9473,
+      "step": 146900
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884869072463327e-05,
+      "loss": 0.94,
+      "step": 146910
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884792907520526e-05,
+      "loss": 0.9666,
+      "step": 146920
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8847167425777265e-05,
+      "loss": 0.8616,
+      "step": 146930
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884640577634926e-05,
+      "loss": 0.9022,
+      "step": 146940
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884564412692126e-05,
+      "loss": 0.9848,
+      "step": 146950
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884488247749326e-05,
+      "loss": 1.0344,
+      "step": 146960
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884412082806526e-05,
+      "loss": 0.8905,
+      "step": 146970
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884335917863726e-05,
+      "loss": 0.9408,
+      "step": 146980
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8842597529209257e-05,
+      "loss": 0.9485,
+      "step": 146990
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884183587978126e-05,
+      "loss": 1.0103,
+      "step": 147000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.884107423035325e-05,
+      "loss": 1.0014,
+      "step": 147010
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8840312580925256e-05,
+      "loss": 0.8877,
+      "step": 147020
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.883955093149725e-05,
+      "loss": 0.9146,
+      "step": 147030
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.883878928206925e-05,
+      "loss": 0.9664,
+      "step": 147040
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.883802763264125e-05,
+      "loss": 0.8823,
+      "step": 147050
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.883726598321325e-05,
+      "loss": 1.0331,
+      "step": 147060
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.883650433378524e-05,
+      "loss": 0.9572,
+      "step": 147070
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.883574268435725e-05,
+      "loss": 0.959,
+      "step": 147080
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8834981034929245e-05,
+      "loss": 0.8651,
+      "step": 147090
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.883421938550124e-05,
+      "loss": 0.9422,
+      "step": 147100
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.883345773607324e-05,
+      "loss": 0.9448,
+      "step": 147110
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8832696086645244e-05,
+      "loss": 1.0053,
+      "step": 147120
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8831934437217235e-05,
+      "loss": 0.9666,
+      "step": 147130
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.883117278778924e-05,
+      "loss": 0.9579,
+      "step": 147140
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8830411138361236e-05,
+      "loss": 0.9917,
+      "step": 147150
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8829649488933234e-05,
+      "loss": 0.9515,
+      "step": 147160
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.882888783950524e-05,
+      "loss": 0.9322,
+      "step": 147170
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8828126190077235e-05,
+      "loss": 0.8629,
+      "step": 147180
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.882736454064923e-05,
+      "loss": 0.9608,
+      "step": 147190
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.882660289122123e-05,
+      "loss": 0.8757,
+      "step": 147200
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8825841241793235e-05,
+      "loss": 0.9075,
+      "step": 147210
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8825079592365225e-05,
+      "loss": 0.8814,
+      "step": 147220
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.882431794293723e-05,
+      "loss": 1.0258,
+      "step": 147230
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.882355629350923e-05,
+      "loss": 1.0823,
+      "step": 147240
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8822794644081225e-05,
+      "loss": 0.8293,
+      "step": 147250
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.882203299465322e-05,
+      "loss": 0.8425,
+      "step": 147260
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8821271345225226e-05,
+      "loss": 0.9137,
+      "step": 147270
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.882050969579722e-05,
+      "loss": 0.9705,
+      "step": 147280
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881974804636922e-05,
+      "loss": 0.9193,
+      "step": 147290
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881898639694122e-05,
+      "loss": 0.9162,
+      "step": 147300
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8818224747513216e-05,
+      "loss": 1.0137,
+      "step": 147310
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8817463098085214e-05,
+      "loss": 0.8492,
+      "step": 147320
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881670144865721e-05,
+      "loss": 0.929,
+      "step": 147330
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881593979922921e-05,
+      "loss": 0.9433,
+      "step": 147340
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881517814980121e-05,
+      "loss": 0.9054,
+      "step": 147350
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881441650037321e-05,
+      "loss": 0.8532,
+      "step": 147360
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881365485094521e-05,
+      "loss": 0.9048,
+      "step": 147370
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881289320151721e-05,
+      "loss": 0.8082,
+      "step": 147380
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.88121315520892e-05,
+      "loss": 0.9421,
+      "step": 147390
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.881136990266121e-05,
+      "loss": 0.9852,
+      "step": 147400
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8810608253233204e-05,
+      "loss": 0.9193,
+      "step": 147410
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.88098466038052e-05,
+      "loss": 0.9383,
+      "step": 147420
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.88090849543772e-05,
+      "loss": 0.8706,
+      "step": 147430
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8808323304949204e-05,
+      "loss": 0.9774,
+      "step": 147440
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8807561655521194e-05,
+      "loss": 0.8645,
+      "step": 147450
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.88068000060932e-05,
+      "loss": 0.9077,
+      "step": 147460
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8806038356665196e-05,
+      "loss": 0.8928,
+      "step": 147470
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.880527670723719e-05,
+      "loss": 0.8699,
+      "step": 147480
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.880451505780919e-05,
+      "loss": 0.9953,
+      "step": 147490
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8803753408381195e-05,
+      "loss": 0.9534,
+      "step": 147500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8802991758953186e-05,
+      "loss": 0.9247,
+      "step": 147510
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.880223010952519e-05,
+      "loss": 0.9149,
+      "step": 147520
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.880146846009719e-05,
+      "loss": 0.9563,
+      "step": 147530
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8800706810669185e-05,
+      "loss": 1.0382,
+      "step": 147540
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879994516124118e-05,
+      "loss": 1.0082,
+      "step": 147550
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879918351181319e-05,
+      "loss": 1.0888,
+      "step": 147560
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8798421862385184e-05,
+      "loss": 0.9742,
+      "step": 147570
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879766021295718e-05,
+      "loss": 0.8962,
+      "step": 147580
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8796898563529186e-05,
+      "loss": 0.959,
+      "step": 147590
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8796136914101177e-05,
+      "loss": 1.0741,
+      "step": 147600
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879537526467318e-05,
+      "loss": 0.9732,
+      "step": 147610
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879461361524518e-05,
+      "loss": 1.0323,
+      "step": 147620
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8793851965817176e-05,
+      "loss": 0.8339,
+      "step": 147630
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879309031638917e-05,
+      "loss": 0.9565,
+      "step": 147640
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879232866696118e-05,
+      "loss": 0.9738,
+      "step": 147650
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879156701753317e-05,
+      "loss": 0.8218,
+      "step": 147660
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879080536810517e-05,
+      "loss": 0.8822,
+      "step": 147670
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.879004371867717e-05,
+      "loss": 0.9402,
+      "step": 147680
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.878928206924917e-05,
+      "loss": 0.8547,
+      "step": 147690
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8788520419821165e-05,
+      "loss": 1.0067,
+      "step": 147700
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.878775877039317e-05,
+      "loss": 0.8896,
+      "step": 147710
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878699712096516e-05,
+      "loss": 0.9294,
+      "step": 147720
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8786235471537164e-05,
+      "loss": 0.8795,
+      "step": 147730
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878547382210916e-05,
+      "loss": 0.9734,
+      "step": 147740
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878471217268116e-05,
+      "loss": 0.8591,
+      "step": 147750
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8783950523253156e-05,
+      "loss": 0.99,
+      "step": 147760
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878318887382516e-05,
+      "loss": 0.8889,
+      "step": 147770
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878242722439716e-05,
+      "loss": 0.9073,
+      "step": 147780
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8781665574969156e-05,
+      "loss": 0.9822,
+      "step": 147790
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878090392554115e-05,
+      "loss": 0.8728,
+      "step": 147800
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.878014227611315e-05,
+      "loss": 1.1102,
+      "step": 147810
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8779380626685155e-05,
+      "loss": 1.0027,
+      "step": 147820
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8778618977257145e-05,
+      "loss": 1.0016,
+      "step": 147830
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.877785732782915e-05,
+      "loss": 0.8873,
+      "step": 147840
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.877709567840115e-05,
+      "loss": 0.9419,
+      "step": 147850
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8776334028973145e-05,
+      "loss": 0.8892,
+      "step": 147860
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.877557237954514e-05,
+      "loss": 0.9305,
+      "step": 147870
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8774810730117146e-05,
+      "loss": 0.9887,
+      "step": 147880
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.877404908068914e-05,
+      "loss": 1.0175,
+      "step": 147890
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.877328743126114e-05,
+      "loss": 1.0008,
+      "step": 147900
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.877252578183314e-05,
+      "loss": 0.8899,
+      "step": 147910
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8771764132405136e-05,
+      "loss": 0.9565,
+      "step": 147920
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8771002482977134e-05,
+      "loss": 0.971,
+      "step": 147930
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.877024083354914e-05,
+      "loss": 0.9264,
+      "step": 147940
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.8769479184121135e-05,
+      "loss": 0.8815,
+      "step": 147950
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876871753469313e-05,
+      "loss": 0.947,
+      "step": 147960
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876795588526514e-05,
+      "loss": 0.9065,
+      "step": 147970
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876719423583713e-05,
+      "loss": 1.0072,
+      "step": 147980
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876643258640913e-05,
+      "loss": 0.8515,
+      "step": 147990
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876567093698113e-05,
+      "loss": 0.9382,
+      "step": 148000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876490928755313e-05,
+      "loss": 1.0649,
+      "step": 148010
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.8764147638125124e-05,
+      "loss": 0.9356,
+      "step": 148020
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876338598869713e-05,
+      "loss": 0.848,
+      "step": 148030
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876262433926912e-05,
+      "loss": 0.9257,
+      "step": 148040
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.8761862689841124e-05,
+      "loss": 0.9422,
+      "step": 148050
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876110104041312e-05,
+      "loss": 0.9684,
+      "step": 148060
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.876033939098512e-05,
+      "loss": 0.9208,
+      "step": 148070
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.8759577741557116e-05,
+      "loss": 0.8898,
+      "step": 148080
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.875881609212912e-05,
+      "loss": 1.0091,
+      "step": 148090
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.875805444270111e-05,
+      "loss": 0.983,
+      "step": 148100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.8757292793273115e-05,
+      "loss": 0.8296,
+      "step": 148110
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.875653114384511e-05,
+      "loss": 0.9196,
+      "step": 148120
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.875576949441711e-05,
+      "loss": 0.9497,
+      "step": 148130
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.875500784498911e-05,
+      "loss": 0.9395,
+      "step": 148140
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.875424619556111e-05,
+      "loss": 0.9442,
+      "step": 148150
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.875348454613311e-05,
+      "loss": 1.0188,
+      "step": 148160
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.875272289670511e-05,
+      "loss": 0.9045,
+      "step": 148170
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.875196124727711e-05,
+      "loss": 0.8098,
+      "step": 148180
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.87511995978491e-05,
+      "loss": 0.9662,
+      "step": 148190
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.8750437948421106e-05,
+      "loss": 0.9835,
+      "step": 148200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.87496762989931e-05,
+      "loss": 0.9036,
+      "step": 148210
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.87489146495651e-05,
+      "loss": 0.9362,
+      "step": 148220
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.87481530001371e-05,
+      "loss": 0.9217,
+      "step": 148230
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.87473913507091e-05,
+      "loss": 0.9585,
+      "step": 148240
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.874662970128109e-05,
+      "loss": 0.9845,
+      "step": 148250
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.87458680518531e-05,
+      "loss": 1.0409,
+      "step": 148260
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.874510640242509e-05,
+      "loss": 0.9813,
+      "step": 148270
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.874434475299709e-05,
+      "loss": 0.9735,
+      "step": 148280
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.874358310356909e-05,
+      "loss": 1.0113,
+      "step": 148290
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.874282145414109e-05,
+      "loss": 0.8925,
+      "step": 148300
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.8742059804713085e-05,
+      "loss": 1.0068,
+      "step": 148310
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.874129815528509e-05,
+      "loss": 0.8739,
+      "step": 148320
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.8740536505857086e-05,
+      "loss": 0.97,
+      "step": 148330
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.8739774856429084e-05,
+      "loss": 0.9353,
+      "step": 148340
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.873901320700108e-05,
+      "loss": 0.9464,
+      "step": 148350
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.873825155757308e-05,
+      "loss": 0.8497,
+      "step": 148360
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.873748990814508e-05,
+      "loss": 0.9434,
+      "step": 148370
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873672825871708e-05,
+      "loss": 0.8649,
+      "step": 148380
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873596660928908e-05,
+      "loss": 0.8629,
+      "step": 148390
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.8735204959861076e-05,
+      "loss": 1.0265,
+      "step": 148400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873444331043308e-05,
+      "loss": 0.967,
+      "step": 148410
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873368166100507e-05,
+      "loss": 1.0125,
+      "step": 148420
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.8732920011577075e-05,
+      "loss": 0.9212,
+      "step": 148430
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873215836214907e-05,
+      "loss": 0.991,
+      "step": 148440
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873139671272107e-05,
+      "loss": 0.9119,
+      "step": 148450
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.873063506329307e-05,
+      "loss": 0.8288,
+      "step": 148460
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872987341386507e-05,
+      "loss": 0.9126,
+      "step": 148470
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872911176443706e-05,
+      "loss": 0.9515,
+      "step": 148480
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.8728350115009066e-05,
+      "loss": 1.025,
+      "step": 148490
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.8727588465581064e-05,
+      "loss": 0.9397,
+      "step": 148500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872682681615306e-05,
+      "loss": 1.0912,
+      "step": 148510
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872606516672506e-05,
+      "loss": 0.9096,
+      "step": 148520
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872530351729706e-05,
+      "loss": 0.8979,
+      "step": 148530
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872454186786906e-05,
+      "loss": 1.0129,
+      "step": 148540
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872378021844106e-05,
+      "loss": 0.9299,
+      "step": 148550
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.8723018569013055e-05,
+      "loss": 1.0408,
+      "step": 148560
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872225691958505e-05,
+      "loss": 0.9438,
+      "step": 148570
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.872149527015706e-05,
+      "loss": 0.8737,
+      "step": 148580
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8720733620729054e-05,
+      "loss": 0.9522,
+      "step": 148590
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871997197130105e-05,
+      "loss": 1.1003,
+      "step": 148600
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871921032187305e-05,
+      "loss": 0.9182,
+      "step": 148610
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8718448672445054e-05,
+      "loss": 0.9423,
+      "step": 148620
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8717687023017044e-05,
+      "loss": 0.8779,
+      "step": 148630
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871692537358905e-05,
+      "loss": 0.9054,
+      "step": 148640
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8716163724161046e-05,
+      "loss": 0.9848,
+      "step": 148650
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8715402074733044e-05,
+      "loss": 0.9737,
+      "step": 148660
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871464042530504e-05,
+      "loss": 0.9326,
+      "step": 148670
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8713878775877045e-05,
+      "loss": 0.9483,
+      "step": 148680
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8713117126449036e-05,
+      "loss": 0.9923,
+      "step": 148690
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871235547702104e-05,
+      "loss": 1.1471,
+      "step": 148700
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871159382759304e-05,
+      "loss": 1.0235,
+      "step": 148710
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8710832178165035e-05,
+      "loss": 1.1087,
+      "step": 148720
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.871007052873703e-05,
+      "loss": 0.9358,
+      "step": 148730
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.870930887930903e-05,
+      "loss": 1.0392,
+      "step": 148740
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8708547229881034e-05,
+      "loss": 0.863,
+      "step": 148750
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.870778558045303e-05,
+      "loss": 0.8457,
+      "step": 148760
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.870702393102503e-05,
+      "loss": 1.0195,
+      "step": 148770
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.870626228159703e-05,
+      "loss": 1.001,
+      "step": 148780
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.870550063216903e-05,
+      "loss": 1.0194,
+      "step": 148790
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.870473898274102e-05,
+      "loss": 0.8485,
+      "step": 148800
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8703977333313026e-05,
+      "loss": 1.0146,
+      "step": 148810
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.870321568388502e-05,
+      "loss": 1.035,
+      "step": 148820
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.870245403445702e-05,
+      "loss": 0.9551,
+      "step": 148830
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.870169238502902e-05,
+      "loss": 0.8776,
+      "step": 148840
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.870093073560102e-05,
+      "loss": 0.8889,
+      "step": 148850
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.870016908617301e-05,
+      "loss": 1.1517,
+      "step": 148860
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.869940743674502e-05,
+      "loss": 0.8963,
+      "step": 148870
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8698645787317015e-05,
+      "loss": 0.9061,
+      "step": 148880
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.869788413788901e-05,
+      "loss": 0.9657,
+      "step": 148890
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.869712248846101e-05,
+      "loss": 1.018,
+      "step": 148900
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8696360839033014e-05,
+      "loss": 0.8383,
+      "step": 148910
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.869559918960501e-05,
+      "loss": 0.8186,
+      "step": 148920
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.869483754017701e-05,
+      "loss": 0.9417,
+      "step": 148930
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8694075890749006e-05,
+      "loss": 1.0523,
+      "step": 148940
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8693314241321004e-05,
+      "loss": 0.9472,
+      "step": 148950
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.869255259189301e-05,
+      "loss": 0.8589,
+      "step": 148960
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8691790942465006e-05,
+      "loss": 0.9445,
+      "step": 148970
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8691029293037e-05,
+      "loss": 0.928,
+      "step": 148980
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8690267643609e-05,
+      "loss": 0.9778,
+      "step": 148990
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8689505994181005e-05,
+      "loss": 1.0249,
+      "step": 149000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8688744344752996e-05,
+      "loss": 0.8543,
+      "step": 149010
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8687982695325e-05,
+      "loss": 0.8978,
+      "step": 149020
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8687221045897e-05,
+      "loss": 0.9798,
+      "step": 149030
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8686459396468995e-05,
+      "loss": 0.9614,
+      "step": 149040
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.868569774704099e-05,
+      "loss": 0.959,
+      "step": 149050
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8684936097612996e-05,
+      "loss": 0.9919,
+      "step": 149060
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.868417444818499e-05,
+      "loss": 0.9646,
+      "step": 149070
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.868341279875699e-05,
+      "loss": 0.9515,
+      "step": 149080
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.868265114932899e-05,
+      "loss": 0.8792,
+      "step": 149090
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8681889499900986e-05,
+      "loss": 0.8632,
+      "step": 149100
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8681127850472984e-05,
+      "loss": 0.8178,
+      "step": 149110
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.868036620104499e-05,
+      "loss": 0.8533,
+      "step": 149120
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8679604551616985e-05,
+      "loss": 1.0284,
+      "step": 149130
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867884290218898e-05,
+      "loss": 0.9535,
+      "step": 149140
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867808125276098e-05,
+      "loss": 0.9727,
+      "step": 149150
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867731960333298e-05,
+      "loss": 1.0009,
+      "step": 149160
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867655795390498e-05,
+      "loss": 0.963,
+      "step": 149170
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867579630447697e-05,
+      "loss": 1.039,
+      "step": 149180
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867503465504898e-05,
+      "loss": 0.8986,
+      "step": 149190
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8674273005620975e-05,
+      "loss": 0.9561,
+      "step": 149200
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867351135619297e-05,
+      "loss": 1.0054,
+      "step": 149210
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.867274970676497e-05,
+      "loss": 0.937,
+      "step": 149220
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8671988057336974e-05,
+      "loss": 0.8814,
+      "step": 149230
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.8671226407908964e-05,
+      "loss": 0.9309,
+      "step": 149240
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.867046475848097e-05,
+      "loss": 1.0435,
+      "step": 149250
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8669703109052966e-05,
+      "loss": 0.8817,
+      "step": 149260
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8668941459624964e-05,
+      "loss": 0.9271,
+      "step": 149270
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866817981019696e-05,
+      "loss": 0.9152,
+      "step": 149280
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8667418160768965e-05,
+      "loss": 0.9049,
+      "step": 149290
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866665651134096e-05,
+      "loss": 1.0169,
+      "step": 149300
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866589486191296e-05,
+      "loss": 0.9364,
+      "step": 149310
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866513321248496e-05,
+      "loss": 0.9197,
+      "step": 149320
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8664371563056955e-05,
+      "loss": 0.9966,
+      "step": 149330
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866360991362896e-05,
+      "loss": 0.8954,
+      "step": 149340
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866284826420096e-05,
+      "loss": 0.9845,
+      "step": 149350
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8662086614772954e-05,
+      "loss": 0.8742,
+      "step": 149360
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.866132496534495e-05,
+      "loss": 0.9217,
+      "step": 149370
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8660563315916956e-05,
+      "loss": 0.9619,
+      "step": 149380
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865980166648895e-05,
+      "loss": 0.9184,
+      "step": 149390
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865904001706095e-05,
+      "loss": 0.8299,
+      "step": 149400
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865827836763295e-05,
+      "loss": 0.9077,
+      "step": 149410
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8657516718204946e-05,
+      "loss": 0.9118,
+      "step": 149420
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865675506877694e-05,
+      "loss": 0.889,
+      "step": 149430
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865599341934895e-05,
+      "loss": 0.8931,
+      "step": 149440
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865523176992094e-05,
+      "loss": 0.8211,
+      "step": 149450
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.865447012049294e-05,
+      "loss": 0.8205,
+      "step": 149460
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.865370847106494e-05,
+      "loss": 0.9716,
+      "step": 149470
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.865294682163694e-05,
+      "loss": 0.8633,
+      "step": 149480
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8652185172208935e-05,
+      "loss": 0.9339,
+      "step": 149490
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.865142352278094e-05,
+      "loss": 0.9249,
+      "step": 149500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.865066187335294e-05,
+      "loss": 1.048,
+      "step": 149510
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8649900223924934e-05,
+      "loss": 1.0184,
+      "step": 149520
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864913857449693e-05,
+      "loss": 1.0012,
+      "step": 149530
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864837692506893e-05,
+      "loss": 0.9539,
+      "step": 149540
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864761527564093e-05,
+      "loss": 0.902,
+      "step": 149550
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864685362621293e-05,
+      "loss": 0.848,
+      "step": 149560
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864609197678493e-05,
+      "loss": 0.8927,
+      "step": 149570
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8645330327356926e-05,
+      "loss": 0.8929,
+      "step": 149580
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864456867792893e-05,
+      "loss": 0.9837,
+      "step": 149590
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864380702850092e-05,
+      "loss": 0.8258,
+      "step": 149600
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8643045379072925e-05,
+      "loss": 0.9539,
+      "step": 149610
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864228372964492e-05,
+      "loss": 0.9943,
+      "step": 149620
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864152208021692e-05,
+      "loss": 0.958,
+      "step": 149630
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.864076043078892e-05,
+      "loss": 0.9821,
+      "step": 149640
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8639998781360915e-05,
+      "loss": 1.1203,
+      "step": 149650
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.863923713193291e-05,
+      "loss": 0.951,
+      "step": 149660
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8638475482504916e-05,
+      "loss": 0.871,
+      "step": 149670
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8637713833076914e-05,
+      "loss": 0.9166,
+      "step": 149680
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.863695218364891e-05,
+      "loss": 0.9115,
+      "step": 149690
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.863619053422091e-05,
+      "loss": 0.9255,
+      "step": 149700
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8635428884792906e-05,
+      "loss": 0.9309,
+      "step": 149710
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.863466723536491e-05,
+      "loss": 0.9288,
+      "step": 149720
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.863390558593691e-05,
+      "loss": 1.0477,
+      "step": 149730
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8633143936508905e-05,
+      "loss": 0.8144,
+      "step": 149740
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.86323822870809e-05,
+      "loss": 0.9681,
+      "step": 149750
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.863162063765291e-05,
+      "loss": 0.9912,
+      "step": 149760
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.86308589882249e-05,
+      "loss": 0.862,
+      "step": 149770
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.86300973387969e-05,
+      "loss": 1.0017,
+      "step": 149780
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.86293356893689e-05,
+      "loss": 0.9872,
+      "step": 149790
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.86285740399409e-05,
+      "loss": 0.9221,
+      "step": 149800
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8627812390512895e-05,
+      "loss": 0.8989,
+      "step": 149810
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.86270507410849e-05,
+      "loss": 1.0367,
+      "step": 149820
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.862628909165689e-05,
+      "loss": 0.9157,
+      "step": 149830
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8625527442228894e-05,
+      "loss": 0.8933,
+      "step": 149840
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.862476579280089e-05,
+      "loss": 0.9317,
+      "step": 149850
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.862400414337289e-05,
+      "loss": 1.0298,
+      "step": 149860
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8623242493944886e-05,
+      "loss": 0.8809,
+      "step": 149870
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.862248084451689e-05,
+      "loss": 0.9859,
+      "step": 149880
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.862171919508889e-05,
+      "loss": 0.9591,
+      "step": 149890
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8620957545660885e-05,
+      "loss": 1.0747,
+      "step": 149900
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.862019589623288e-05,
+      "loss": 0.9832,
+      "step": 149910
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861943424680488e-05,
+      "loss": 1.0373,
+      "step": 149920
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.8618672597376884e-05,
+      "loss": 0.9907,
+      "step": 149930
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861791094794888e-05,
+      "loss": 0.9191,
+      "step": 149940
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861714929852088e-05,
+      "loss": 0.8849,
+      "step": 149950
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861638764909288e-05,
+      "loss": 0.9326,
+      "step": 149960
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861562599966488e-05,
+      "loss": 0.8929,
+      "step": 149970
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861486435023687e-05,
+      "loss": 0.982,
+      "step": 149980
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.8614102700808876e-05,
+      "loss": 0.9498,
+      "step": 149990
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.8613341051380873e-05,
+      "loss": 0.9999,
+      "step": 150000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861257940195287e-05,
+      "loss": 0.8999,
+      "step": 150010
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861181775252487e-05,
+      "loss": 0.9276,
+      "step": 150020
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861105610309687e-05,
+      "loss": 1.0566,
+      "step": 150030
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.861029445366886e-05,
+      "loss": 0.8642,
+      "step": 150040
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.860953280424087e-05,
+      "loss": 0.8713,
+      "step": 150050
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.8608771154812865e-05,
+      "loss": 0.9391,
+      "step": 150060
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.860800950538486e-05,
+      "loss": 0.9262,
+      "step": 150070
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.860724785595686e-05,
+      "loss": 1.0058,
+      "step": 150080
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.8606486206528864e-05,
+      "loss": 0.9634,
+      "step": 150090
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.860572455710086e-05,
+      "loss": 0.9439,
+      "step": 150100
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.860496290767286e-05,
+      "loss": 1.0073,
+      "step": 150110
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.860420125824486e-05,
+      "loss": 0.9411,
+      "step": 150120
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8603439608816854e-05,
+      "loss": 0.9604,
+      "step": 150130
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.860267795938886e-05,
+      "loss": 0.9297,
+      "step": 150140
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.860191630996085e-05,
+      "loss": 0.8674,
+      "step": 150150
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.860115466053285e-05,
+      "loss": 1.0269,
+      "step": 150160
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.860039301110485e-05,
+      "loss": 0.8643,
+      "step": 150170
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859963136167685e-05,
+      "loss": 0.92,
+      "step": 150180
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8598869712248846e-05,
+      "loss": 0.9115,
+      "step": 150190
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859810806282085e-05,
+      "loss": 0.9198,
+      "step": 150200
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859734641339284e-05,
+      "loss": 0.9563,
+      "step": 150210
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8596584763964845e-05,
+      "loss": 0.9501,
+      "step": 150220
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859582311453684e-05,
+      "loss": 0.9458,
+      "step": 150230
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859506146510884e-05,
+      "loss": 0.998,
+      "step": 150240
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859429981568084e-05,
+      "loss": 0.8692,
+      "step": 150250
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859353816625284e-05,
+      "loss": 0.9703,
+      "step": 150260
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859277651682484e-05,
+      "loss": 1.0921,
+      "step": 150270
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8592014867396836e-05,
+      "loss": 0.9344,
+      "step": 150280
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8591253217968834e-05,
+      "loss": 0.9648,
+      "step": 150290
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.859049156854083e-05,
+      "loss": 0.9461,
+      "step": 150300
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8589729919112836e-05,
+      "loss": 1.0634,
+      "step": 150310
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.858896826968483e-05,
+      "loss": 0.9046,
+      "step": 150320
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.858820662025683e-05,
+      "loss": 0.9755,
+      "step": 150330
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.858744497082883e-05,
+      "loss": 0.9192,
+      "step": 150340
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.858668332140083e-05,
+      "loss": 0.9343,
+      "step": 150350
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.858592167197282e-05,
+      "loss": 0.7976,
+      "step": 150360
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.858516002254483e-05,
+      "loss": 0.9047,
+      "step": 150370
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8584398373116825e-05,
+      "loss": 0.8781,
+      "step": 150380
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.858363672368882e-05,
+      "loss": 0.9004,
+      "step": 150390
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.858287507426082e-05,
+      "loss": 0.8957,
+      "step": 150400
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8582113424832824e-05,
+      "loss": 0.8902,
+      "step": 150410
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8581351775404815e-05,
+      "loss": 0.8949,
+      "step": 150420
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.858059012597682e-05,
+      "loss": 0.9705,
+      "step": 150430
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8579828476548816e-05,
+      "loss": 0.9715,
+      "step": 150440
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8579066827120814e-05,
+      "loss": 1.0317,
+      "step": 150450
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.857830517769281e-05,
+      "loss": 0.8493,
+      "step": 150460
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8577543528264815e-05,
+      "loss": 0.9032,
+      "step": 150470
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.857678187883681e-05,
+      "loss": 0.9538,
+      "step": 150480
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.857602022940881e-05,
+      "loss": 0.8105,
+      "step": 150490
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.857525857998081e-05,
+      "loss": 0.9662,
+      "step": 150500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8574496930552805e-05,
+      "loss": 0.9482,
+      "step": 150510
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.857373528112481e-05,
+      "loss": 0.8773,
+      "step": 150520
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.857297363169681e-05,
+      "loss": 0.902,
+      "step": 150530
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8572211982268804e-05,
+      "loss": 0.9879,
+      "step": 150540
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.85714503328408e-05,
+      "loss": 0.9972,
+      "step": 150550
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.85706886834128e-05,
+      "loss": 0.9358,
+      "step": 150560
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.85699270339848e-05,
+      "loss": 0.9372,
+      "step": 150570
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.85691653845568e-05,
+      "loss": 0.9137,
+      "step": 150580
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856840373512879e-05,
+      "loss": 0.946,
+      "step": 150590
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8567642085700796e-05,
+      "loss": 0.9208,
+      "step": 150600
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8566880436272793e-05,
+      "loss": 1.03,
+      "step": 150610
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856611878684479e-05,
+      "loss": 0.8357,
+      "step": 150620
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856535713741679e-05,
+      "loss": 0.9782,
+      "step": 150630
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856459548798879e-05,
+      "loss": 0.9367,
+      "step": 150640
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856383383856078e-05,
+      "loss": 0.9157,
+      "step": 150650
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856307218913279e-05,
+      "loss": 0.9128,
+      "step": 150660
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8562310539704785e-05,
+      "loss": 0.9549,
+      "step": 150670
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856154889027678e-05,
+      "loss": 0.9553,
+      "step": 150680
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.856078724084879e-05,
+      "loss": 0.899,
+      "step": 150690
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8560025591420784e-05,
+      "loss": 0.8621,
+      "step": 150700
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.855926394199278e-05,
+      "loss": 0.9562,
+      "step": 150710
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.855850229256478e-05,
+      "loss": 0.9155,
+      "step": 150720
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8557740643136783e-05,
+      "loss": 0.9086,
+      "step": 150730
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8556978993708774e-05,
+      "loss": 0.9337,
+      "step": 150740
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.855621734428078e-05,
+      "loss": 0.8826,
+      "step": 150750
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8555455694852776e-05,
+      "loss": 0.9596,
+      "step": 150760
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.855469404542477e-05,
+      "loss": 0.9228,
+      "step": 150770
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.855393239599677e-05,
+      "loss": 0.9573,
+      "step": 150780
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8553170746568775e-05,
+      "loss": 0.9234,
+      "step": 150790
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8552409097140766e-05,
+      "loss": 0.9768,
+      "step": 150800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.855164744771277e-05,
+      "loss": 0.8726,
+      "step": 150810
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.855088579828477e-05,
+      "loss": 0.984,
+      "step": 150820
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8550124148856765e-05,
+      "loss": 0.8065,
+      "step": 150830
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854936249942876e-05,
+      "loss": 0.8786,
+      "step": 150840
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8548600850000767e-05,
+      "loss": 0.9012,
+      "step": 150850
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854783920057276e-05,
+      "loss": 1.0003,
+      "step": 150860
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854707755114476e-05,
+      "loss": 0.9337,
+      "step": 150870
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854631590171676e-05,
+      "loss": 0.9511,
+      "step": 150880
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8545554252288756e-05,
+      "loss": 0.7949,
+      "step": 150890
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854479260286076e-05,
+      "loss": 0.918,
+      "step": 150900
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854403095343276e-05,
+      "loss": 1.0192,
+      "step": 150910
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8543269304004756e-05,
+      "loss": 0.9375,
+      "step": 150920
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854250765457675e-05,
+      "loss": 0.9021,
+      "step": 150930
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854174600514876e-05,
+      "loss": 0.9419,
+      "step": 150940
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854098435572075e-05,
+      "loss": 0.8518,
+      "step": 150950
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.854022270629275e-05,
+      "loss": 0.9092,
+      "step": 150960
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.853946105686475e-05,
+      "loss": 0.9626,
+      "step": 150970
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.853869940743675e-05,
+      "loss": 0.9236,
+      "step": 150980
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.8537937758008745e-05,
+      "loss": 0.789,
+      "step": 150990
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.853717610858075e-05,
+      "loss": 0.8656,
+      "step": 151000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.853641445915274e-05,
+      "loss": 1.0771,
+      "step": 151010
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8535652809724744e-05,
+      "loss": 0.8091,
+      "step": 151020
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8534891160296735e-05,
+      "loss": 0.8672,
+      "step": 151030
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.853412951086874e-05,
+      "loss": 0.9034,
+      "step": 151040
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8533367861440736e-05,
+      "loss": 0.9025,
+      "step": 151050
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8532606212012734e-05,
+      "loss": 0.8609,
+      "step": 151060
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.853184456258474e-05,
+      "loss": 0.9704,
+      "step": 151070
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8531082913156735e-05,
+      "loss": 0.8726,
+      "step": 151080
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.853032126372873e-05,
+      "loss": 0.9237,
+      "step": 151090
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852955961430073e-05,
+      "loss": 0.9868,
+      "step": 151100
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8528797964872735e-05,
+      "loss": 0.9131,
+      "step": 151110
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8528036315444725e-05,
+      "loss": 0.9399,
+      "step": 151120
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852727466601673e-05,
+      "loss": 0.927,
+      "step": 151130
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852651301658873e-05,
+      "loss": 0.8378,
+      "step": 151140
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8525751367160724e-05,
+      "loss": 1.0705,
+      "step": 151150
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852498971773272e-05,
+      "loss": 0.8123,
+      "step": 151160
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8524228068304726e-05,
+      "loss": 0.9499,
+      "step": 151170
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852346641887672e-05,
+      "loss": 0.9734,
+      "step": 151180
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852270476944872e-05,
+      "loss": 0.9547,
+      "step": 151190
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.852194312002072e-05,
+      "loss": 1.1204,
+      "step": 151200
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.8521181470592716e-05,
+      "loss": 0.9745,
+      "step": 151210
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8520419821164714e-05,
+      "loss": 0.9175,
+      "step": 151220
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851965817173672e-05,
+      "loss": 0.9733,
+      "step": 151230
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851889652230871e-05,
+      "loss": 0.9085,
+      "step": 151240
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851813487288071e-05,
+      "loss": 0.9003,
+      "step": 151250
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851737322345271e-05,
+      "loss": 0.9751,
+      "step": 151260
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851661157402471e-05,
+      "loss": 0.9429,
+      "step": 151270
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851584992459671e-05,
+      "loss": 0.8454,
+      "step": 151280
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851508827516871e-05,
+      "loss": 0.8617,
+      "step": 151290
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851432662574071e-05,
+      "loss": 1.0637,
+      "step": 151300
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8513564976312704e-05,
+      "loss": 0.9615,
+      "step": 151310
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.851280332688471e-05,
+      "loss": 1.1008,
+      "step": 151320
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.85120416774567e-05,
+      "loss": 0.881,
+      "step": 151330
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8511280028028703e-05,
+      "loss": 0.9478,
+      "step": 151340
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.85105183786007e-05,
+      "loss": 0.9462,
+      "step": 151350
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.85097567291727e-05,
+      "loss": 0.9428,
+      "step": 151360
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8508995079744696e-05,
+      "loss": 0.8929,
+      "step": 151370
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.85082334303167e-05,
+      "loss": 0.8621,
+      "step": 151380
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.850747178088869e-05,
+      "loss": 0.96,
+      "step": 151390
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8506710131460695e-05,
+      "loss": 0.9416,
+      "step": 151400
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.850594848203269e-05,
+      "loss": 0.8227,
+      "step": 151410
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.850518683260469e-05,
+      "loss": 0.9764,
+      "step": 151420
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.850442518317669e-05,
+      "loss": 0.9074,
+      "step": 151430
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.850366353374869e-05,
+      "loss": 1.0123,
+      "step": 151440
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.850290188432068e-05,
+      "loss": 1.0006,
+      "step": 151450
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8502140234892687e-05,
+      "loss": 0.8929,
+      "step": 151460
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8501378585464684e-05,
+      "loss": 0.9486,
+      "step": 151470
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.850061693603668e-05,
+      "loss": 1.0501,
+      "step": 151480
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8499855286608686e-05,
+      "loss": 1.0078,
+      "step": 151490
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8499093637180676e-05,
+      "loss": 0.9022,
+      "step": 151500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849833198775268e-05,
+      "loss": 0.8571,
+      "step": 151510
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849757033832468e-05,
+      "loss": 0.8796,
+      "step": 151520
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8496808688896676e-05,
+      "loss": 0.9307,
+      "step": 151530
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849604703946867e-05,
+      "loss": 0.8537,
+      "step": 151540
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849528539004068e-05,
+      "loss": 0.8617,
+      "step": 151550
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849452374061267e-05,
+      "loss": 1.0054,
+      "step": 151560
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849376209118467e-05,
+      "loss": 0.8659,
+      "step": 151570
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849300044175667e-05,
+      "loss": 1.0025,
+      "step": 151580
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849223879232867e-05,
+      "loss": 0.964,
+      "step": 151590
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8491477142900665e-05,
+      "loss": 0.9874,
+      "step": 151600
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.849071549347267e-05,
+      "loss": 0.8841,
+      "step": 151610
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.848995384404466e-05,
+      "loss": 0.9725,
+      "step": 151620
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8489192194616664e-05,
+      "loss": 0.906,
+      "step": 151630
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.848843054518866e-05,
+      "loss": 0.8703,
+      "step": 151640
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.848766889576066e-05,
+      "loss": 0.9338,
+      "step": 151650
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848690724633266e-05,
+      "loss": 0.8374,
+      "step": 151660
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848614559690466e-05,
+      "loss": 0.9503,
+      "step": 151670
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848538394747666e-05,
+      "loss": 1.0029,
+      "step": 151680
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8484622298048655e-05,
+      "loss": 0.9075,
+      "step": 151690
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848386064862066e-05,
+      "loss": 0.9365,
+      "step": 151700
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848309899919265e-05,
+      "loss": 0.8351,
+      "step": 151710
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8482337349764655e-05,
+      "loss": 0.9664,
+      "step": 151720
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848157570033665e-05,
+      "loss": 0.9049,
+      "step": 151730
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848081405090865e-05,
+      "loss": 1.004,
+      "step": 151740
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.848005240148065e-05,
+      "loss": 0.8232,
+      "step": 151750
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847929075205265e-05,
+      "loss": 0.9446,
+      "step": 151760
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847852910262464e-05,
+      "loss": 1.0021,
+      "step": 151770
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8477767453196646e-05,
+      "loss": 0.9027,
+      "step": 151780
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8477005803768644e-05,
+      "loss": 0.9085,
+      "step": 151790
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847624415434064e-05,
+      "loss": 0.9122,
+      "step": 151800
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847548250491264e-05,
+      "loss": 0.9392,
+      "step": 151810
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847472085548464e-05,
+      "loss": 0.9667,
+      "step": 151820
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8473959206056634e-05,
+      "loss": 1.0101,
+      "step": 151830
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847319755662864e-05,
+      "loss": 0.9103,
+      "step": 151840
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8472435907200635e-05,
+      "loss": 1.059,
+      "step": 151850
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847167425777263e-05,
+      "loss": 0.8893,
+      "step": 151860
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.847091260834464e-05,
+      "loss": 0.8544,
+      "step": 151870
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8470150958916634e-05,
+      "loss": 0.8172,
+      "step": 151880
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846938930948863e-05,
+      "loss": 0.9576,
+      "step": 151890
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846862766006063e-05,
+      "loss": 0.9927,
+      "step": 151900
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8467866010632634e-05,
+      "loss": 0.9271,
+      "step": 151910
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8467104361204624e-05,
+      "loss": 0.95,
+      "step": 151920
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846634271177663e-05,
+      "loss": 0.9089,
+      "step": 151930
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846558106234862e-05,
+      "loss": 0.836,
+      "step": 151940
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8464819412920623e-05,
+      "loss": 0.9517,
+      "step": 151950
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846405776349262e-05,
+      "loss": 0.9741,
+      "step": 151960
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846329611406462e-05,
+      "loss": 0.8472,
+      "step": 151970
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8462534464636616e-05,
+      "loss": 0.9703,
+      "step": 151980
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846177281520862e-05,
+      "loss": 0.9977,
+      "step": 151990
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.846101116578061e-05,
+      "loss": 0.8938,
+      "step": 152000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8460249516352615e-05,
+      "loss": 0.886,
+      "step": 152010
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.845948786692461e-05,
+      "loss": 1.0277,
+      "step": 152020
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.845872621749661e-05,
+      "loss": 1.0428,
+      "step": 152030
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.845796456806861e-05,
+      "loss": 0.8903,
+      "step": 152040
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.845720291864061e-05,
+      "loss": 0.9857,
+      "step": 152050
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.845644126921261e-05,
+      "loss": 0.7956,
+      "step": 152060
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.8455679619784607e-05,
+      "loss": 0.8389,
+      "step": 152070
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.845491797035661e-05,
+      "loss": 0.9547,
+      "step": 152080
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.84541563209286e-05,
+      "loss": 0.967,
+      "step": 152090
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8453394671500606e-05,
+      "loss": 0.9644,
+      "step": 152100
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.84526330220726e-05,
+      "loss": 0.9139,
+      "step": 152110
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.84518713726446e-05,
+      "loss": 0.8759,
+      "step": 152120
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.84511097232166e-05,
+      "loss": 0.8694,
+      "step": 152130
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.84503480737886e-05,
+      "loss": 0.9816,
+      "step": 152140
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.844958642436059e-05,
+      "loss": 0.8627,
+      "step": 152150
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.84488247749326e-05,
+      "loss": 1.0347,
+      "step": 152160
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8448063125504595e-05,
+      "loss": 0.9079,
+      "step": 152170
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.844730147607659e-05,
+      "loss": 0.9144,
+      "step": 152180
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.844653982664859e-05,
+      "loss": 0.9864,
+      "step": 152190
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8445778177220594e-05,
+      "loss": 0.946,
+      "step": 152200
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8445016527792585e-05,
+      "loss": 0.9084,
+      "step": 152210
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.844425487836459e-05,
+      "loss": 0.9448,
+      "step": 152220
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8443493228936586e-05,
+      "loss": 0.896,
+      "step": 152230
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8442731579508584e-05,
+      "loss": 1.0474,
+      "step": 152240
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.844196993008058e-05,
+      "loss": 0.9695,
+      "step": 152250
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8441208280652586e-05,
+      "loss": 1.054,
+      "step": 152260
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.844044663122458e-05,
+      "loss": 0.9217,
+      "step": 152270
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.843968498179658e-05,
+      "loss": 0.9976,
+      "step": 152280
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8438923332368585e-05,
+      "loss": 0.923,
+      "step": 152290
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.8438161682940575e-05,
+      "loss": 0.8586,
+      "step": 152300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.843740003351258e-05,
+      "loss": 0.9171,
+      "step": 152310
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.843663838408458e-05,
+      "loss": 0.8989,
+      "step": 152320
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8435876734656575e-05,
+      "loss": 0.9973,
+      "step": 152330
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.843511508522857e-05,
+      "loss": 0.993,
+      "step": 152340
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8434353435800576e-05,
+      "loss": 0.9885,
+      "step": 152350
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.843359178637257e-05,
+      "loss": 0.8348,
+      "step": 152360
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.843283013694457e-05,
+      "loss": 0.9889,
+      "step": 152370
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.843206848751657e-05,
+      "loss": 0.9345,
+      "step": 152380
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8431306838088566e-05,
+      "loss": 0.9202,
+      "step": 152390
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8430545188660564e-05,
+      "loss": 0.9863,
+      "step": 152400
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842978353923256e-05,
+      "loss": 0.9806,
+      "step": 152410
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842902188980456e-05,
+      "loss": 0.8944,
+      "step": 152420
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842826024037656e-05,
+      "loss": 0.9702,
+      "step": 152430
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842749859094856e-05,
+      "loss": 0.9131,
+      "step": 152440
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842673694152056e-05,
+      "loss": 0.9715,
+      "step": 152450
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842597529209256e-05,
+      "loss": 0.963,
+      "step": 152460
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842521364266455e-05,
+      "loss": 1.0169,
+      "step": 152470
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842445199323656e-05,
+      "loss": 0.9415,
+      "step": 152480
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8423690343808554e-05,
+      "loss": 0.9695,
+      "step": 152490
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842292869438055e-05,
+      "loss": 0.87,
+      "step": 152500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.842216704495255e-05,
+      "loss": 0.9589,
+      "step": 152510
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8421405395524554e-05,
+      "loss": 0.9091,
+      "step": 152520
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.8420643746096544e-05,
+      "loss": 0.8357,
+      "step": 152530
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841988209666855e-05,
+      "loss": 0.857,
+      "step": 152540
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8419120447240546e-05,
+      "loss": 0.8518,
+      "step": 152550
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8418358797812543e-05,
+      "loss": 0.9247,
+      "step": 152560
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841759714838454e-05,
+      "loss": 0.916,
+      "step": 152570
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8416835498956545e-05,
+      "loss": 0.8642,
+      "step": 152580
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8416073849528536e-05,
+      "loss": 1.0501,
+      "step": 152590
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841531220010054e-05,
+      "loss": 0.9604,
+      "step": 152600
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841455055067254e-05,
+      "loss": 0.956,
+      "step": 152610
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8413788901244535e-05,
+      "loss": 0.9408,
+      "step": 152620
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841302725181653e-05,
+      "loss": 0.9111,
+      "step": 152630
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841226560238854e-05,
+      "loss": 0.9164,
+      "step": 152640
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8411503952960534e-05,
+      "loss": 0.9165,
+      "step": 152650
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.841074230353253e-05,
+      "loss": 0.9888,
+      "step": 152660
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8409980654104536e-05,
+      "loss": 0.9844,
+      "step": 152670
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840921900467653e-05,
+      "loss": 0.8656,
+      "step": 152680
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840845735524853e-05,
+      "loss": 0.8499,
+      "step": 152690
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840769570582053e-05,
+      "loss": 0.913,
+      "step": 152700
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.8406934056392526e-05,
+      "loss": 0.8826,
+      "step": 152710
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840617240696452e-05,
+      "loss": 0.8743,
+      "step": 152720
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840541075753653e-05,
+      "loss": 0.9509,
+      "step": 152730
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840464910810852e-05,
+      "loss": 0.9832,
+      "step": 152740
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.840388745868052e-05,
+      "loss": 0.8993,
+      "step": 152750
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.840312580925252e-05,
+      "loss": 0.9458,
+      "step": 152760
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.840236415982452e-05,
+      "loss": 0.9562,
+      "step": 152770
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8401602510396515e-05,
+      "loss": 0.9343,
+      "step": 152780
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.840084086096852e-05,
+      "loss": 0.911,
+      "step": 152790
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.840007921154051e-05,
+      "loss": 1.015,
+      "step": 152800
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8399317562112514e-05,
+      "loss": 0.9756,
+      "step": 152810
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.839855591268451e-05,
+      "loss": 0.8914,
+      "step": 152820
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.839779426325651e-05,
+      "loss": 0.9078,
+      "step": 152830
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8397032613828506e-05,
+      "loss": 1.054,
+      "step": 152840
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.839627096440051e-05,
+      "loss": 0.8967,
+      "step": 152850
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.839550931497251e-05,
+      "loss": 0.8303,
+      "step": 152860
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8394747665544506e-05,
+      "loss": 0.8878,
+      "step": 152870
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.83939860161165e-05,
+      "loss": 0.9005,
+      "step": 152880
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.83932243666885e-05,
+      "loss": 0.887,
+      "step": 152890
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8392462717260505e-05,
+      "loss": 0.8924,
+      "step": 152900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8391701067832495e-05,
+      "loss": 0.9045,
+      "step": 152910
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.83909394184045e-05,
+      "loss": 0.9983,
+      "step": 152920
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.83901777689765e-05,
+      "loss": 0.9741,
+      "step": 152930
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8389416119548495e-05,
+      "loss": 0.9284,
+      "step": 152940
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.838865447012049e-05,
+      "loss": 0.9216,
+      "step": 152950
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.8387892820692496e-05,
+      "loss": 0.9034,
+      "step": 152960
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838713117126449e-05,
+      "loss": 0.9136,
+      "step": 152970
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838636952183649e-05,
+      "loss": 0.9233,
+      "step": 152980
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838560787240849e-05,
+      "loss": 0.9555,
+      "step": 152990
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.8384846222980486e-05,
+      "loss": 0.9676,
+      "step": 153000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.8384084573552484e-05,
+      "loss": 0.9922,
+      "step": 153010
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838332292412449e-05,
+      "loss": 0.8438,
+      "step": 153020
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.8382561274696485e-05,
+      "loss": 0.9922,
+      "step": 153030
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838179962526848e-05,
+      "loss": 0.9772,
+      "step": 153040
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838103797584048e-05,
+      "loss": 0.9313,
+      "step": 153050
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.838027632641248e-05,
+      "loss": 0.8281,
+      "step": 153060
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837951467698448e-05,
+      "loss": 0.9958,
+      "step": 153070
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837875302755648e-05,
+      "loss": 0.8932,
+      "step": 153080
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837799137812848e-05,
+      "loss": 0.947,
+      "step": 153090
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.8377229728700474e-05,
+      "loss": 0.87,
+      "step": 153100
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837646807927248e-05,
+      "loss": 1.0214,
+      "step": 153110
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837570642984447e-05,
+      "loss": 0.9578,
+      "step": 153120
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.8374944780416474e-05,
+      "loss": 0.894,
+      "step": 153130
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837418313098847e-05,
+      "loss": 0.8686,
+      "step": 153140
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837342148156047e-05,
+      "loss": 0.9239,
+      "step": 153150
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.8372659832132466e-05,
+      "loss": 0.9401,
+      "step": 153160
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837189818270447e-05,
+      "loss": 0.9503,
+      "step": 153170
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.837113653327646e-05,
+      "loss": 0.958,
+      "step": 153180
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.8370374883848465e-05,
+      "loss": 0.9444,
+      "step": 153190
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836961323442046e-05,
+      "loss": 0.9151,
+      "step": 153200
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836885158499246e-05,
+      "loss": 0.9466,
+      "step": 153210
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836808993556446e-05,
+      "loss": 0.8249,
+      "step": 153220
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836732828613646e-05,
+      "loss": 0.9529,
+      "step": 153230
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836656663670846e-05,
+      "loss": 0.9049,
+      "step": 153240
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836580498728046e-05,
+      "loss": 0.9821,
+      "step": 153250
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836504333785246e-05,
+      "loss": 0.8337,
+      "step": 153260
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836428168842445e-05,
+      "loss": 1.077,
+      "step": 153270
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.8363520038996456e-05,
+      "loss": 0.9121,
+      "step": 153280
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.8362758389568453e-05,
+      "loss": 0.8997,
+      "step": 153290
+    },
+    {
+      "epoch": 7.0,
+      "eval_cer": 0.622565658582502,
+      "eval_em": 0.007473982970671713,
+      "eval_f1": 0.007473982970671713,
+      "eval_loss": 0.9018365144729614,
+      "eval_runtime": 1010.8094,
+      "eval_samples_per_second": 10.457,
+      "eval_steps_per_second": 1.308,
+      "eval_wer": 0.9925260170293283,
+      "step": 153298
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836199674014045e-05,
+      "loss": 0.8928,
+      "step": 153300
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836123509071245e-05,
+      "loss": 0.9085,
+      "step": 153310
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.836047344128445e-05,
+      "loss": 0.912,
+      "step": 153320
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835971179185644e-05,
+      "loss": 0.866,
+      "step": 153330
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835895014242845e-05,
+      "loss": 0.9029,
+      "step": 153340
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835818849300044e-05,
+      "loss": 0.8793,
+      "step": 153350
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835742684357244e-05,
+      "loss": 0.9563,
+      "step": 153360
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835666519414444e-05,
+      "loss": 0.8875,
+      "step": 153370
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835590354471644e-05,
+      "loss": 0.928,
+      "step": 153380
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.8355141895288435e-05,
+      "loss": 0.8867,
+      "step": 153390
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.835438024586044e-05,
+      "loss": 0.966,
+      "step": 153400
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.8353618596432437e-05,
+      "loss": 0.9482,
+      "step": 153410
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.8352856947004434e-05,
+      "loss": 0.8521,
+      "step": 153420
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.835209529757643e-05,
+      "loss": 0.8651,
+      "step": 153430
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.835133364814843e-05,
+      "loss": 0.8983,
+      "step": 153440
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.835057199872043e-05,
+      "loss": 0.8597,
+      "step": 153450
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834981034929243e-05,
+      "loss": 0.8765,
+      "step": 153460
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834904869986443e-05,
+      "loss": 0.8147,
+      "step": 153470
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.8348287050436426e-05,
+      "loss": 0.8886,
+      "step": 153480
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834752540100843e-05,
+      "loss": 0.7875,
+      "step": 153490
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834676375158042e-05,
+      "loss": 0.9334,
+      "step": 153500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.8346002102152425e-05,
+      "loss": 0.8811,
+      "step": 153510
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834524045272442e-05,
+      "loss": 0.9417,
+      "step": 153520
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834447880329642e-05,
+      "loss": 0.9044,
+      "step": 153530
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834371715386842e-05,
+      "loss": 0.8328,
+      "step": 153540
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834295550444042e-05,
+      "loss": 1.0373,
+      "step": 153550
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.834219385501241e-05,
+      "loss": 0.9881,
+      "step": 153560
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.8341432205584416e-05,
+      "loss": 0.8438,
+      "step": 153570
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.8340670556156414e-05,
+      "loss": 0.8859,
+      "step": 153580
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.833990890672841e-05,
+      "loss": 0.96,
+      "step": 153590
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.833914725730041e-05,
+      "loss": 0.8264,
+      "step": 153600
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.833838560787241e-05,
+      "loss": 0.8374,
+      "step": 153610
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.833762395844441e-05,
+      "loss": 0.7738,
+      "step": 153620
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.833686230901641e-05,
+      "loss": 0.9293,
+      "step": 153630
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8336100659588405e-05,
+      "loss": 0.8649,
+      "step": 153640
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.83353390101604e-05,
+      "loss": 0.9603,
+      "step": 153650
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.833457736073241e-05,
+      "loss": 0.863,
+      "step": 153660
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8333815711304405e-05,
+      "loss": 0.9558,
+      "step": 153670
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.83330540618764e-05,
+      "loss": 0.8716,
+      "step": 153680
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.83322924124484e-05,
+      "loss": 0.8245,
+      "step": 153690
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8331530763020404e-05,
+      "loss": 0.9084,
+      "step": 153700
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8330769113592394e-05,
+      "loss": 0.8348,
+      "step": 153710
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.83300074641644e-05,
+      "loss": 0.9531,
+      "step": 153720
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8329245814736396e-05,
+      "loss": 1.0103,
+      "step": 153730
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8328484165308394e-05,
+      "loss": 0.879,
+      "step": 153740
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.832772251588039e-05,
+      "loss": 0.8822,
+      "step": 153750
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8326960866452395e-05,
+      "loss": 0.9287,
+      "step": 153760
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8326199217024386e-05,
+      "loss": 0.8426,
+      "step": 153770
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.832543756759639e-05,
+      "loss": 0.8168,
+      "step": 153780
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.832467591816839e-05,
+      "loss": 0.7903,
+      "step": 153790
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8323914268740385e-05,
+      "loss": 0.9796,
+      "step": 153800
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.832315261931238e-05,
+      "loss": 0.8928,
+      "step": 153810
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.832239096988438e-05,
+      "loss": 0.9881,
+      "step": 153820
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.8321629320456384e-05,
+      "loss": 0.8599,
+      "step": 153830
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.832086767102838e-05,
+      "loss": 0.9276,
+      "step": 153840
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.832010602160038e-05,
+      "loss": 0.982,
+      "step": 153850
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831934437217238e-05,
+      "loss": 0.9104,
+      "step": 153860
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831858272274438e-05,
+      "loss": 0.8613,
+      "step": 153870
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831782107331637e-05,
+      "loss": 0.8569,
+      "step": 153880
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8317059423888376e-05,
+      "loss": 0.9786,
+      "step": 153890
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8316297774460373e-05,
+      "loss": 0.9765,
+      "step": 153900
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831553612503237e-05,
+      "loss": 0.8104,
+      "step": 153910
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831477447560437e-05,
+      "loss": 0.904,
+      "step": 153920
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831401282617637e-05,
+      "loss": 0.9079,
+      "step": 153930
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831325117674836e-05,
+      "loss": 0.9774,
+      "step": 153940
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831248952732037e-05,
+      "loss": 0.8659,
+      "step": 153950
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8311727877892365e-05,
+      "loss": 0.8929,
+      "step": 153960
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831096622846436e-05,
+      "loss": 0.906,
+      "step": 153970
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.831020457903636e-05,
+      "loss": 1.0297,
+      "step": 153980
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8309442929608364e-05,
+      "loss": 0.8471,
+      "step": 153990
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.830868128018036e-05,
+      "loss": 0.8454,
+      "step": 154000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.830791963075236e-05,
+      "loss": 0.877,
+      "step": 154010
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8307157981324357e-05,
+      "loss": 0.9971,
+      "step": 154020
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8306396331896354e-05,
+      "loss": 0.8294,
+      "step": 154030
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.830563468246836e-05,
+      "loss": 0.9295,
+      "step": 154040
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8304873033040356e-05,
+      "loss": 1.0144,
+      "step": 154050
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.830411138361235e-05,
+      "loss": 0.9029,
+      "step": 154060
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.830334973418435e-05,
+      "loss": 0.955,
+      "step": 154070
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8302588084756355e-05,
+      "loss": 0.8805,
+      "step": 154080
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8301826435328346e-05,
+      "loss": 0.8125,
+      "step": 154090
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.830106478590035e-05,
+      "loss": 0.8843,
+      "step": 154100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.830030313647235e-05,
+      "loss": 0.8658,
+      "step": 154110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8299541487044345e-05,
+      "loss": 0.9055,
+      "step": 154120
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829877983761634e-05,
+      "loss": 0.9001,
+      "step": 154130
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8298018188188346e-05,
+      "loss": 0.9066,
+      "step": 154140
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829725653876034e-05,
+      "loss": 0.8806,
+      "step": 154150
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829649488933234e-05,
+      "loss": 0.8121,
+      "step": 154160
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829573323990434e-05,
+      "loss": 0.8576,
+      "step": 154170
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8294971590476336e-05,
+      "loss": 0.9338,
+      "step": 154180
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8294209941048334e-05,
+      "loss": 0.8658,
+      "step": 154190
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829344829162034e-05,
+      "loss": 0.8783,
+      "step": 154200
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8292686642192336e-05,
+      "loss": 0.9772,
+      "step": 154210
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829192499276433e-05,
+      "loss": 0.9605,
+      "step": 154220
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829116334333633e-05,
+      "loss": 0.9139,
+      "step": 154230
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.829040169390833e-05,
+      "loss": 1.018,
+      "step": 154240
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.828964004448033e-05,
+      "loss": 0.9856,
+      "step": 154250
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.828887839505232e-05,
+      "loss": 0.8493,
+      "step": 154260
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.828811674562433e-05,
+      "loss": 0.9455,
+      "step": 154270
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.8287355096196325e-05,
+      "loss": 0.9154,
+      "step": 154280
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.828659344676832e-05,
+      "loss": 0.937,
+      "step": 154290
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.828583179734032e-05,
+      "loss": 0.9367,
+      "step": 154300
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8285070147912324e-05,
+      "loss": 0.9015,
+      "step": 154310
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8284308498484314e-05,
+      "loss": 0.9017,
+      "step": 154320
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.828354684905632e-05,
+      "loss": 0.8411,
+      "step": 154330
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8282785199628316e-05,
+      "loss": 0.9035,
+      "step": 154340
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8282023550200314e-05,
+      "loss": 0.8578,
+      "step": 154350
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.828126190077231e-05,
+      "loss": 0.7618,
+      "step": 154360
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8280500251344315e-05,
+      "loss": 0.8854,
+      "step": 154370
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.827973860191631e-05,
+      "loss": 0.8927,
+      "step": 154380
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.827897695248831e-05,
+      "loss": 0.951,
+      "step": 154390
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.827821530306031e-05,
+      "loss": 0.8729,
+      "step": 154400
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8277453653632305e-05,
+      "loss": 0.8004,
+      "step": 154410
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.827669200420431e-05,
+      "loss": 0.8848,
+      "step": 154420
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.827593035477631e-05,
+      "loss": 0.8693,
+      "step": 154430
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8275168705348304e-05,
+      "loss": 0.9138,
+      "step": 154440
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.82744070559203e-05,
+      "loss": 0.9011,
+      "step": 154450
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8273645406492306e-05,
+      "loss": 0.8723,
+      "step": 154460
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.82728837570643e-05,
+      "loss": 0.9281,
+      "step": 154470
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.82721221076363e-05,
+      "loss": 0.9802,
+      "step": 154480
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.82713604582083e-05,
+      "loss": 0.7861,
+      "step": 154490
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.8270598808780296e-05,
+      "loss": 0.9456,
+      "step": 154500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.8269837159352293e-05,
+      "loss": 0.8764,
+      "step": 154510
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.82690755099243e-05,
+      "loss": 0.8865,
+      "step": 154520
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826831386049629e-05,
+      "loss": 0.9461,
+      "step": 154530
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826755221106829e-05,
+      "loss": 0.845,
+      "step": 154540
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826679056164029e-05,
+      "loss": 0.986,
+      "step": 154550
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826602891221229e-05,
+      "loss": 0.9728,
+      "step": 154560
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.8265267262784285e-05,
+      "loss": 0.9104,
+      "step": 154570
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826450561335629e-05,
+      "loss": 0.9423,
+      "step": 154580
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826374396392829e-05,
+      "loss": 0.8999,
+      "step": 154590
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.8262982314500284e-05,
+      "loss": 0.8302,
+      "step": 154600
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826222066507228e-05,
+      "loss": 0.8793,
+      "step": 154610
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826145901564428e-05,
+      "loss": 0.8815,
+      "step": 154620
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.826069736621628e-05,
+      "loss": 0.8867,
+      "step": 154630
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825993571678828e-05,
+      "loss": 0.8155,
+      "step": 154640
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825917406736028e-05,
+      "loss": 0.9323,
+      "step": 154650
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.8258412417932276e-05,
+      "loss": 0.8936,
+      "step": 154660
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825765076850428e-05,
+      "loss": 0.9475,
+      "step": 154670
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825688911907627e-05,
+      "loss": 1.0154,
+      "step": 154680
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.8256127469648275e-05,
+      "loss": 0.86,
+      "step": 154690
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825536582022027e-05,
+      "loss": 0.9598,
+      "step": 154700
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825460417079227e-05,
+      "loss": 0.9423,
+      "step": 154710
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.825384252136427e-05,
+      "loss": 0.9802,
+      "step": 154720
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8253080871936265e-05,
+      "loss": 0.8514,
+      "step": 154730
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.825231922250826e-05,
+      "loss": 0.9671,
+      "step": 154740
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8251557573080267e-05,
+      "loss": 0.9487,
+      "step": 154750
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8250795923652264e-05,
+      "loss": 0.8799,
+      "step": 154760
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.825003427422426e-05,
+      "loss": 0.8611,
+      "step": 154770
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824927262479626e-05,
+      "loss": 0.8058,
+      "step": 154780
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8248510975368256e-05,
+      "loss": 0.971,
+      "step": 154790
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824774932594026e-05,
+      "loss": 0.9334,
+      "step": 154800
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824698767651226e-05,
+      "loss": 0.9073,
+      "step": 154810
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8246226027084256e-05,
+      "loss": 1.0237,
+      "step": 154820
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824546437765625e-05,
+      "loss": 0.8858,
+      "step": 154830
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824470272822826e-05,
+      "loss": 0.968,
+      "step": 154840
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824394107880025e-05,
+      "loss": 1.0109,
+      "step": 154850
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824317942937225e-05,
+      "loss": 0.7899,
+      "step": 154860
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824241777994425e-05,
+      "loss": 1.0387,
+      "step": 154870
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824165613051625e-05,
+      "loss": 1.0083,
+      "step": 154880
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8240894481088245e-05,
+      "loss": 0.8915,
+      "step": 154890
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.824013283166025e-05,
+      "loss": 0.9204,
+      "step": 154900
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.823937118223224e-05,
+      "loss": 0.9015,
+      "step": 154910
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.8238609532804244e-05,
+      "loss": 0.9952,
+      "step": 154920
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.823784788337624e-05,
+      "loss": 0.9659,
+      "step": 154930
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.823708623394824e-05,
+      "loss": 0.9399,
+      "step": 154940
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8236324584520236e-05,
+      "loss": 0.9153,
+      "step": 154950
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.823556293509224e-05,
+      "loss": 0.9066,
+      "step": 154960
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.823480128566424e-05,
+      "loss": 0.8568,
+      "step": 154970
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8234039636236235e-05,
+      "loss": 0.8211,
+      "step": 154980
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.823327798680823e-05,
+      "loss": 0.9988,
+      "step": 154990
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.823251633738023e-05,
+      "loss": 0.9291,
+      "step": 155000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8231754687952235e-05,
+      "loss": 0.9249,
+      "step": 155010
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.823099303852423e-05,
+      "loss": 0.9782,
+      "step": 155020
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.823023138909623e-05,
+      "loss": 0.7889,
+      "step": 155030
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822946973966823e-05,
+      "loss": 0.8648,
+      "step": 155040
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822870809024023e-05,
+      "loss": 0.9665,
+      "step": 155050
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822794644081222e-05,
+      "loss": 0.9198,
+      "step": 155060
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8227184791384226e-05,
+      "loss": 0.9122,
+      "step": 155070
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8226423141956224e-05,
+      "loss": 0.9576,
+      "step": 155080
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822566149252822e-05,
+      "loss": 0.8975,
+      "step": 155090
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822489984310022e-05,
+      "loss": 0.8648,
+      "step": 155100
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822413819367222e-05,
+      "loss": 0.9291,
+      "step": 155110
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8223376544244213e-05,
+      "loss": 0.9195,
+      "step": 155120
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822261489481622e-05,
+      "loss": 0.8709,
+      "step": 155130
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.8221853245388215e-05,
+      "loss": 1.0533,
+      "step": 155140
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.822109159596021e-05,
+      "loss": 0.9407,
+      "step": 155150
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.822032994653221e-05,
+      "loss": 0.9095,
+      "step": 155160
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.821956829710421e-05,
+      "loss": 0.8867,
+      "step": 155170
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.821880664767621e-05,
+      "loss": 0.9278,
+      "step": 155180
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.821804499824821e-05,
+      "loss": 0.9598,
+      "step": 155190
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.821728334882021e-05,
+      "loss": 0.9804,
+      "step": 155200
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.8216521699392204e-05,
+      "loss": 0.9294,
+      "step": 155210
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.821576004996421e-05,
+      "loss": 0.8116,
+      "step": 155220
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.82149984005362e-05,
+      "loss": 1.0019,
+      "step": 155230
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.82142367511082e-05,
+      "loss": 0.912,
+      "step": 155240
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.82134751016802e-05,
+      "loss": 0.8518,
+      "step": 155250
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.82127134522522e-05,
+      "loss": 0.8969,
+      "step": 155260
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.8211951802824196e-05,
+      "loss": 0.8553,
+      "step": 155270
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.82111901533962e-05,
+      "loss": 0.9052,
+      "step": 155280
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.821042850396819e-05,
+      "loss": 1.0221,
+      "step": 155290
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.8209666854540195e-05,
+      "loss": 0.9453,
+      "step": 155300
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.820890520511219e-05,
+      "loss": 0.8816,
+      "step": 155310
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.820814355568419e-05,
+      "loss": 0.8844,
+      "step": 155320
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.820738190625619e-05,
+      "loss": 0.9211,
+      "step": 155330
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.820662025682819e-05,
+      "loss": 0.9882,
+      "step": 155340
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.820585860740018e-05,
+      "loss": 0.8793,
+      "step": 155350
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.8205096957972187e-05,
+      "loss": 0.9399,
+      "step": 155360
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.8204335308544184e-05,
+      "loss": 1.0027,
+      "step": 155370
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.820357365911618e-05,
+      "loss": 0.9996,
+      "step": 155380
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8202812009688186e-05,
+      "loss": 0.9725,
+      "step": 155390
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.820205036026018e-05,
+      "loss": 0.8527,
+      "step": 155400
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.820128871083218e-05,
+      "loss": 0.8691,
+      "step": 155410
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.820052706140418e-05,
+      "loss": 0.922,
+      "step": 155420
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819976541197618e-05,
+      "loss": 0.8597,
+      "step": 155430
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819900376254817e-05,
+      "loss": 0.8852,
+      "step": 155440
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819824211312018e-05,
+      "loss": 0.7753,
+      "step": 155450
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8197480463692175e-05,
+      "loss": 0.9647,
+      "step": 155460
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819671881426417e-05,
+      "loss": 1.0077,
+      "step": 155470
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819595716483617e-05,
+      "loss": 0.9023,
+      "step": 155480
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8195195515408174e-05,
+      "loss": 0.9248,
+      "step": 155490
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8194433865980165e-05,
+      "loss": 0.9659,
+      "step": 155500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819367221655217e-05,
+      "loss": 0.9541,
+      "step": 155510
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8192910567124166e-05,
+      "loss": 0.861,
+      "step": 155520
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8192148917696164e-05,
+      "loss": 0.9859,
+      "step": 155530
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.819138726826816e-05,
+      "loss": 0.9004,
+      "step": 155540
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8190625618840165e-05,
+      "loss": 0.9248,
+      "step": 155550
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.818986396941216e-05,
+      "loss": 0.8691,
+      "step": 155560
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.818910231998416e-05,
+      "loss": 0.8664,
+      "step": 155570
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.818834067055616e-05,
+      "loss": 0.9045,
+      "step": 155580
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.8187579021128155e-05,
+      "loss": 0.8721,
+      "step": 155590
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818681737170016e-05,
+      "loss": 0.9142,
+      "step": 155600
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818605572227216e-05,
+      "loss": 1.0,
+      "step": 155610
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8185294072844155e-05,
+      "loss": 0.9483,
+      "step": 155620
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818453242341615e-05,
+      "loss": 1.0603,
+      "step": 155630
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818377077398815e-05,
+      "loss": 0.9082,
+      "step": 155640
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818300912456015e-05,
+      "loss": 0.9749,
+      "step": 155650
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818224747513215e-05,
+      "loss": 1.0715,
+      "step": 155660
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.818148582570414e-05,
+      "loss": 0.8904,
+      "step": 155670
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8180724176276146e-05,
+      "loss": 0.9043,
+      "step": 155680
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8179962526848144e-05,
+      "loss": 0.8005,
+      "step": 155690
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817920087742014e-05,
+      "loss": 0.9292,
+      "step": 155700
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817843922799214e-05,
+      "loss": 0.9505,
+      "step": 155710
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817767757856414e-05,
+      "loss": 1.0026,
+      "step": 155720
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8176915929136133e-05,
+      "loss": 0.9239,
+      "step": 155730
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817615427970814e-05,
+      "loss": 0.8614,
+      "step": 155740
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8175392630280135e-05,
+      "loss": 0.9355,
+      "step": 155750
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817463098085213e-05,
+      "loss": 0.9561,
+      "step": 155760
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817386933142414e-05,
+      "loss": 0.8782,
+      "step": 155770
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8173107681996134e-05,
+      "loss": 0.9145,
+      "step": 155780
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817234603256813e-05,
+      "loss": 0.9296,
+      "step": 155790
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.817158438314013e-05,
+      "loss": 0.9026,
+      "step": 155800
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.8170822733712134e-05,
+      "loss": 0.9076,
+      "step": 155810
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8170061084284124e-05,
+      "loss": 0.9245,
+      "step": 155820
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816929943485613e-05,
+      "loss": 0.904,
+      "step": 155830
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8168537785428126e-05,
+      "loss": 0.8248,
+      "step": 155840
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816777613600012e-05,
+      "loss": 0.88,
+      "step": 155850
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816701448657212e-05,
+      "loss": 0.8957,
+      "step": 155860
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8166252837144125e-05,
+      "loss": 0.8623,
+      "step": 155870
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8165491187716116e-05,
+      "loss": 0.9394,
+      "step": 155880
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816472953828812e-05,
+      "loss": 0.9055,
+      "step": 155890
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816396788886012e-05,
+      "loss": 0.8261,
+      "step": 155900
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8163206239432115e-05,
+      "loss": 1.0312,
+      "step": 155910
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816244459000411e-05,
+      "loss": 0.9378,
+      "step": 155920
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816168294057612e-05,
+      "loss": 0.9731,
+      "step": 155930
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816092129114811e-05,
+      "loss": 1.0145,
+      "step": 155940
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.816015964172011e-05,
+      "loss": 0.971,
+      "step": 155950
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.815939799229211e-05,
+      "loss": 0.8788,
+      "step": 155960
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8158636342864107e-05,
+      "loss": 0.8645,
+      "step": 155970
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.815787469343611e-05,
+      "loss": 1.0624,
+      "step": 155980
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.815711304400811e-05,
+      "loss": 0.9117,
+      "step": 155990
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8156351394580106e-05,
+      "loss": 1.0014,
+      "step": 156000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.81555897451521e-05,
+      "loss": 0.9084,
+      "step": 156010
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.815482809572411e-05,
+      "loss": 0.8781,
+      "step": 156020
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.81540664462961e-05,
+      "loss": 0.9758,
+      "step": 156030
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.81533047968681e-05,
+      "loss": 0.9129,
+      "step": 156040
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.81525431474401e-05,
+      "loss": 0.928,
+      "step": 156050
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.81517814980121e-05,
+      "loss": 0.901,
+      "step": 156060
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8151019848584095e-05,
+      "loss": 0.8415,
+      "step": 156070
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.81502581991561e-05,
+      "loss": 0.9989,
+      "step": 156080
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.814949654972809e-05,
+      "loss": 0.8752,
+      "step": 156090
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8148734900300094e-05,
+      "loss": 0.9208,
+      "step": 156100
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8147973250872085e-05,
+      "loss": 0.9976,
+      "step": 156110
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.814721160144409e-05,
+      "loss": 0.9227,
+      "step": 156120
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8146449952016086e-05,
+      "loss": 0.9683,
+      "step": 156130
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8145688302588084e-05,
+      "loss": 1.0337,
+      "step": 156140
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.814492665316008e-05,
+      "loss": 0.8945,
+      "step": 156150
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8144165003732086e-05,
+      "loss": 0.9624,
+      "step": 156160
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.814340335430408e-05,
+      "loss": 1.0359,
+      "step": 156170
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.814264170487608e-05,
+      "loss": 0.8535,
+      "step": 156180
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8141880055448085e-05,
+      "loss": 0.9838,
+      "step": 156190
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8141118406020075e-05,
+      "loss": 0.9226,
+      "step": 156200
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.814035675659208e-05,
+      "loss": 0.8972,
+      "step": 156210
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.813959510716408e-05,
+      "loss": 0.9136,
+      "step": 156220
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8138833457736075e-05,
+      "loss": 0.8436,
+      "step": 156230
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.813807180830807e-05,
+      "loss": 0.9473,
+      "step": 156240
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8137310158880076e-05,
+      "loss": 0.8451,
+      "step": 156250
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813654850945207e-05,
+      "loss": 0.8073,
+      "step": 156260
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813578686002407e-05,
+      "loss": 0.913,
+      "step": 156270
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813502521059607e-05,
+      "loss": 0.9495,
+      "step": 156280
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8134263561168066e-05,
+      "loss": 0.9038,
+      "step": 156290
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8133501911740064e-05,
+      "loss": 0.8723,
+      "step": 156300
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813274026231207e-05,
+      "loss": 0.7936,
+      "step": 156310
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813197861288406e-05,
+      "loss": 1.0262,
+      "step": 156320
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813121696345606e-05,
+      "loss": 1.0858,
+      "step": 156330
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.813045531402806e-05,
+      "loss": 0.8762,
+      "step": 156340
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812969366460006e-05,
+      "loss": 0.9145,
+      "step": 156350
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812893201517206e-05,
+      "loss": 0.8928,
+      "step": 156360
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812817036574406e-05,
+      "loss": 0.8924,
+      "step": 156370
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812740871631606e-05,
+      "loss": 0.8633,
+      "step": 156380
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8126647066888054e-05,
+      "loss": 0.9715,
+      "step": 156390
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812588541746006e-05,
+      "loss": 0.8778,
+      "step": 156400
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812512376803205e-05,
+      "loss": 0.7834,
+      "step": 156410
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8124362118604054e-05,
+      "loss": 0.9644,
+      "step": 156420
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812360046917605e-05,
+      "loss": 0.8543,
+      "step": 156430
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812283881974805e-05,
+      "loss": 0.8634,
+      "step": 156440
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8122077170320046e-05,
+      "loss": 1.0076,
+      "step": 156450
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812131552089205e-05,
+      "loss": 1.0458,
+      "step": 156460
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.812055387146404e-05,
+      "loss": 0.8345,
+      "step": 156470
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8119792222036045e-05,
+      "loss": 0.8519,
+      "step": 156480
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811903057260804e-05,
+      "loss": 0.8883,
+      "step": 156490
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811826892318004e-05,
+      "loss": 0.9363,
+      "step": 156500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811750727375204e-05,
+      "loss": 0.8874,
+      "step": 156510
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811674562432404e-05,
+      "loss": 0.9478,
+      "step": 156520
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811598397489603e-05,
+      "loss": 0.9391,
+      "step": 156530
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811522232546804e-05,
+      "loss": 0.908,
+      "step": 156540
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8114460676040034e-05,
+      "loss": 0.8237,
+      "step": 156550
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811369902661203e-05,
+      "loss": 0.8462,
+      "step": 156560
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8112937377184036e-05,
+      "loss": 0.9462,
+      "step": 156570
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8112175727756027e-05,
+      "loss": 0.9686,
+      "step": 156580
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811141407832803e-05,
+      "loss": 1.0253,
+      "step": 156590
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.811065242890003e-05,
+      "loss": 0.8784,
+      "step": 156600
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8109890779472026e-05,
+      "loss": 0.9043,
+      "step": 156610
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810912913004402e-05,
+      "loss": 0.8776,
+      "step": 156620
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810836748061603e-05,
+      "loss": 0.8944,
+      "step": 156630
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810760583118802e-05,
+      "loss": 0.8374,
+      "step": 156640
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810684418176002e-05,
+      "loss": 1.0369,
+      "step": 156650
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810608253233202e-05,
+      "loss": 0.9338,
+      "step": 156660
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810532088290402e-05,
+      "loss": 0.9754,
+      "step": 156670
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8104559233476015e-05,
+      "loss": 0.8852,
+      "step": 156680
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.810379758404802e-05,
+      "loss": 0.9401,
+      "step": 156690
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.810303593462001e-05,
+      "loss": 0.992,
+      "step": 156700
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8102274285192014e-05,
+      "loss": 0.8668,
+      "step": 156710
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.810151263576401e-05,
+      "loss": 0.8689,
+      "step": 156720
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.810075098633601e-05,
+      "loss": 0.9474,
+      "step": 156730
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8099989336908006e-05,
+      "loss": 0.9192,
+      "step": 156740
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.809922768748001e-05,
+      "loss": 0.9192,
+      "step": 156750
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.809846603805201e-05,
+      "loss": 0.9623,
+      "step": 156760
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8097704388624006e-05,
+      "loss": 0.9486,
+      "step": 156770
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.809694273919601e-05,
+      "loss": 0.8531,
+      "step": 156780
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8096181089768e-05,
+      "loss": 0.8704,
+      "step": 156790
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8095419440340005e-05,
+      "loss": 0.9172,
+      "step": 156800
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8094657790912e-05,
+      "loss": 0.8229,
+      "step": 156810
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8093896141484e-05,
+      "loss": 0.9281,
+      "step": 156820
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8093134492056e-05,
+      "loss": 0.9229,
+      "step": 156830
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8092372842628e-05,
+      "loss": 0.8762,
+      "step": 156840
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.809161119319999e-05,
+      "loss": 0.9646,
+      "step": 156850
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8090849543771996e-05,
+      "loss": 0.9656,
+      "step": 156860
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8090087894343994e-05,
+      "loss": 0.917,
+      "step": 156870
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.808932624491599e-05,
+      "loss": 0.9373,
+      "step": 156880
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.808856459548799e-05,
+      "loss": 0.8548,
+      "step": 156890
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.808780294605999e-05,
+      "loss": 0.9593,
+      "step": 156900
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.8087041296631984e-05,
+      "loss": 0.9152,
+      "step": 156910
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.808627964720399e-05,
+      "loss": 0.8686,
+      "step": 156920
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8085517997775985e-05,
+      "loss": 0.9116,
+      "step": 156930
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.808475634834798e-05,
+      "loss": 0.9173,
+      "step": 156940
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.808399469891998e-05,
+      "loss": 0.9767,
+      "step": 156950
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8083233049491984e-05,
+      "loss": 0.9325,
+      "step": 156960
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.808247140006398e-05,
+      "loss": 0.9478,
+      "step": 156970
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.808170975063598e-05,
+      "loss": 0.9625,
+      "step": 156980
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8080948101207984e-05,
+      "loss": 0.8283,
+      "step": 156990
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8080186451779974e-05,
+      "loss": 0.8671,
+      "step": 157000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807942480235198e-05,
+      "loss": 0.8853,
+      "step": 157010
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807866315292397e-05,
+      "loss": 0.8532,
+      "step": 157020
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8077901503495974e-05,
+      "loss": 0.9548,
+      "step": 157030
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807713985406797e-05,
+      "loss": 0.9388,
+      "step": 157040
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807637820463997e-05,
+      "loss": 0.7906,
+      "step": 157050
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8075616555211966e-05,
+      "loss": 0.9627,
+      "step": 157060
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807485490578397e-05,
+      "loss": 0.9077,
+      "step": 157070
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807409325635596e-05,
+      "loss": 0.9001,
+      "step": 157080
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.8073331606927965e-05,
+      "loss": 0.9217,
+      "step": 157090
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807256995749996e-05,
+      "loss": 0.9475,
+      "step": 157100
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807180830807196e-05,
+      "loss": 0.9007,
+      "step": 157110
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807104665864396e-05,
+      "loss": 1.0254,
+      "step": 157120
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.807028500921596e-05,
+      "loss": 0.9273,
+      "step": 157130
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806952335978796e-05,
+      "loss": 1.1516,
+      "step": 157140
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806876171035996e-05,
+      "loss": 0.9478,
+      "step": 157150
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806800006093196e-05,
+      "loss": 0.903,
+      "step": 157160
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806723841150395e-05,
+      "loss": 1.0085,
+      "step": 157170
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8066476762075956e-05,
+      "loss": 0.8845,
+      "step": 157180
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806571511264795e-05,
+      "loss": 0.8093,
+      "step": 157190
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806495346321995e-05,
+      "loss": 0.964,
+      "step": 157200
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806419181379195e-05,
+      "loss": 1.017,
+      "step": 157210
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806343016436395e-05,
+      "loss": 0.8772,
+      "step": 157220
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806266851493594e-05,
+      "loss": 1.0023,
+      "step": 157230
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806190686550795e-05,
+      "loss": 0.8484,
+      "step": 157240
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8061145216079945e-05,
+      "loss": 0.8738,
+      "step": 157250
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.806038356665194e-05,
+      "loss": 0.9193,
+      "step": 157260
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.805962191722394e-05,
+      "loss": 0.9168,
+      "step": 157270
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8058860267795944e-05,
+      "loss": 0.9544,
+      "step": 157280
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8058098618367935e-05,
+      "loss": 0.9484,
+      "step": 157290
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.805733696893994e-05,
+      "loss": 0.8896,
+      "step": 157300
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8056575319511936e-05,
+      "loss": 0.9828,
+      "step": 157310
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8055813670083934e-05,
+      "loss": 0.9372,
+      "step": 157320
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.805505202065593e-05,
+      "loss": 1.0196,
+      "step": 157330
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.8054290371227936e-05,
+      "loss": 0.9287,
+      "step": 157340
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.805352872179993e-05,
+      "loss": 0.8662,
+      "step": 157350
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.805276707237193e-05,
+      "loss": 0.8285,
+      "step": 157360
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8052005422943935e-05,
+      "loss": 0.9477,
+      "step": 157370
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8051243773515926e-05,
+      "loss": 0.9598,
+      "step": 157380
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.805048212408793e-05,
+      "loss": 0.8729,
+      "step": 157390
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804972047465993e-05,
+      "loss": 0.7405,
+      "step": 157400
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8048958825231925e-05,
+      "loss": 0.9573,
+      "step": 157410
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804819717580392e-05,
+      "loss": 0.9707,
+      "step": 157420
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8047435526375926e-05,
+      "loss": 0.825,
+      "step": 157430
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804667387694792e-05,
+      "loss": 0.8959,
+      "step": 157440
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804591222751992e-05,
+      "loss": 0.9749,
+      "step": 157450
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804515057809192e-05,
+      "loss": 0.9486,
+      "step": 157460
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8044388928663916e-05,
+      "loss": 0.9592,
+      "step": 157470
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8043627279235914e-05,
+      "loss": 0.8694,
+      "step": 157480
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804286562980791e-05,
+      "loss": 0.9899,
+      "step": 157490
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804210398037991e-05,
+      "loss": 0.9596,
+      "step": 157500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804134233095191e-05,
+      "loss": 1.0265,
+      "step": 157510
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.804058068152391e-05,
+      "loss": 0.8947,
+      "step": 157520
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.803981903209591e-05,
+      "loss": 0.895,
+      "step": 157530
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8039057382667905e-05,
+      "loss": 0.875,
+      "step": 157540
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.80382957332399e-05,
+      "loss": 0.8905,
+      "step": 157550
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.803753408381191e-05,
+      "loss": 0.9804,
+      "step": 157560
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8036772434383904e-05,
+      "loss": 0.8858,
+      "step": 157570
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.80360107849559e-05,
+      "loss": 0.9943,
+      "step": 157580
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.80352491355279e-05,
+      "loss": 0.856,
+      "step": 157590
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8034487486099904e-05,
+      "loss": 0.9603,
+      "step": 157600
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8033725836671894e-05,
+      "loss": 0.9082,
+      "step": 157610
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.80329641872439e-05,
+      "loss": 0.9842,
+      "step": 157620
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8032202537815896e-05,
+      "loss": 0.9111,
+      "step": 157630
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8031440888387894e-05,
+      "loss": 0.8505,
+      "step": 157640
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.803067923895989e-05,
+      "loss": 0.8982,
+      "step": 157650
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8029917589531895e-05,
+      "loss": 0.823,
+      "step": 157660
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8029155940103886e-05,
+      "loss": 0.9063,
+      "step": 157670
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802839429067589e-05,
+      "loss": 0.8273,
+      "step": 157680
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802763264124789e-05,
+      "loss": 0.8277,
+      "step": 157690
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8026870991819885e-05,
+      "loss": 0.9503,
+      "step": 157700
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802610934239188e-05,
+      "loss": 0.9952,
+      "step": 157710
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802534769296389e-05,
+      "loss": 0.9445,
+      "step": 157720
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8024586043535884e-05,
+      "loss": 1.0045,
+      "step": 157730
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802382439410788e-05,
+      "loss": 0.94,
+      "step": 157740
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802306274467988e-05,
+      "loss": 0.9797,
+      "step": 157750
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802230109525188e-05,
+      "loss": 0.9326,
+      "step": 157760
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802153944582388e-05,
+      "loss": 0.9879,
+      "step": 157770
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.802077779639588e-05,
+      "loss": 0.8872,
+      "step": 157780
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8020016146967876e-05,
+      "loss": 0.8997,
+      "step": 157790
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801925449753987e-05,
+      "loss": 0.9394,
+      "step": 157800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801849284811188e-05,
+      "loss": 0.8995,
+      "step": 157810
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801773119868387e-05,
+      "loss": 0.9961,
+      "step": 157820
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801696954925587e-05,
+      "loss": 0.8902,
+      "step": 157830
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801620789982787e-05,
+      "loss": 0.8736,
+      "step": 157840
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801544625039987e-05,
+      "loss": 0.968,
+      "step": 157850
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8014684600971865e-05,
+      "loss": 0.8585,
+      "step": 157860
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801392295154387e-05,
+      "loss": 0.9777,
+      "step": 157870
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801316130211586e-05,
+      "loss": 0.9439,
+      "step": 157880
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8012399652687864e-05,
+      "loss": 0.9489,
+      "step": 157890
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801163800325986e-05,
+      "loss": 0.9085,
+      "step": 157900
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.801087635383186e-05,
+      "loss": 0.9352,
+      "step": 157910
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8010114704403856e-05,
+      "loss": 0.9314,
+      "step": 157920
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.800935305497586e-05,
+      "loss": 0.847,
+      "step": 157930
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.800859140554786e-05,
+      "loss": 0.9276,
+      "step": 157940
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8007829756119856e-05,
+      "loss": 0.909,
+      "step": 157950
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.800706810669185e-05,
+      "loss": 0.8903,
+      "step": 157960
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.800630645726385e-05,
+      "loss": 0.9254,
+      "step": 157970
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8005544807835855e-05,
+      "loss": 0.8989,
+      "step": 157980
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8004783158407846e-05,
+      "loss": 1.0068,
+      "step": 157990
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.800402150897985e-05,
+      "loss": 0.9127,
+      "step": 158000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.800325985955185e-05,
+      "loss": 0.8815,
+      "step": 158010
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.8002498210123845e-05,
+      "loss": 0.8252,
+      "step": 158020
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.800173656069584e-05,
+      "loss": 1.1548,
+      "step": 158030
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.8000974911267846e-05,
+      "loss": 0.87,
+      "step": 158040
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.800021326183984e-05,
+      "loss": 0.907,
+      "step": 158050
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799945161241184e-05,
+      "loss": 0.8688,
+      "step": 158060
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799868996298384e-05,
+      "loss": 0.8395,
+      "step": 158070
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.7997928313555836e-05,
+      "loss": 0.9435,
+      "step": 158080
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.7997166664127834e-05,
+      "loss": 0.8884,
+      "step": 158090
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799640501469984e-05,
+      "loss": 0.9946,
+      "step": 158100
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.7995643365271835e-05,
+      "loss": 0.8944,
+      "step": 158110
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799488171584383e-05,
+      "loss": 0.9606,
+      "step": 158120
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799412006641583e-05,
+      "loss": 0.9057,
+      "step": 158130
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799335841698783e-05,
+      "loss": 0.9525,
+      "step": 158140
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799259676755983e-05,
+      "loss": 0.9565,
+      "step": 158150
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799183511813183e-05,
+      "loss": 0.9662,
+      "step": 158160
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.799107346870383e-05,
+      "loss": 0.9086,
+      "step": 158170
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.7990311819275825e-05,
+      "loss": 0.9336,
+      "step": 158180
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.798955016984783e-05,
+      "loss": 0.9971,
+      "step": 158190
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.798878852041982e-05,
+      "loss": 1.0012,
+      "step": 158200
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.7988026870991824e-05,
+      "loss": 0.9148,
+      "step": 158210
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.798726522156382e-05,
+      "loss": 1.0442,
+      "step": 158220
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798650357213582e-05,
+      "loss": 0.9105,
+      "step": 158230
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7985741922707816e-05,
+      "loss": 0.896,
+      "step": 158240
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798498027327982e-05,
+      "loss": 1.0778,
+      "step": 158250
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798421862385181e-05,
+      "loss": 0.9047,
+      "step": 158260
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7983456974423815e-05,
+      "loss": 0.9397,
+      "step": 158270
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798269532499581e-05,
+      "loss": 0.9164,
+      "step": 158280
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798193367556781e-05,
+      "loss": 0.9281,
+      "step": 158290
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798117202613981e-05,
+      "loss": 0.9088,
+      "step": 158300
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.798041037671181e-05,
+      "loss": 0.8727,
+      "step": 158310
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.797964872728381e-05,
+      "loss": 0.9601,
+      "step": 158320
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.797888707785581e-05,
+      "loss": 0.905,
+      "step": 158330
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7978125428427804e-05,
+      "loss": 0.8725,
+      "step": 158340
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.79773637789998e-05,
+      "loss": 0.7243,
+      "step": 158350
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7976602129571806e-05,
+      "loss": 0.8748,
+      "step": 158360
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7975840480143803e-05,
+      "loss": 0.9562,
+      "step": 158370
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.79750788307158e-05,
+      "loss": 0.8988,
+      "step": 158380
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.79743171812878e-05,
+      "loss": 0.9424,
+      "step": 158390
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7973555531859796e-05,
+      "loss": 1.0377,
+      "step": 158400
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.797279388243179e-05,
+      "loss": 0.9059,
+      "step": 158410
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.79720322330038e-05,
+      "loss": 1.0568,
+      "step": 158420
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.797127058357579e-05,
+      "loss": 0.8797,
+      "step": 158430
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.797050893414779e-05,
+      "loss": 0.9751,
+      "step": 158440
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796974728471979e-05,
+      "loss": 0.8461,
+      "step": 158450
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796898563529179e-05,
+      "loss": 0.902,
+      "step": 158460
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7968223985863785e-05,
+      "loss": 0.8499,
+      "step": 158470
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796746233643579e-05,
+      "loss": 0.9488,
+      "step": 158480
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796670068700779e-05,
+      "loss": 0.8833,
+      "step": 158490
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7965939037579784e-05,
+      "loss": 0.9273,
+      "step": 158500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796517738815178e-05,
+      "loss": 0.8005,
+      "step": 158510
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796441573872378e-05,
+      "loss": 0.8403,
+      "step": 158520
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796365408929578e-05,
+      "loss": 0.8768,
+      "step": 158530
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796289243986778e-05,
+      "loss": 0.8468,
+      "step": 158540
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796213079043978e-05,
+      "loss": 0.9045,
+      "step": 158550
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7961369141011776e-05,
+      "loss": 0.8887,
+      "step": 158560
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.796060749158378e-05,
+      "loss": 0.8949,
+      "step": 158570
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.795984584215577e-05,
+      "loss": 0.9686,
+      "step": 158580
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7959084192727775e-05,
+      "loss": 0.8712,
+      "step": 158590
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.795832254329977e-05,
+      "loss": 0.9028,
+      "step": 158600
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.795756089387177e-05,
+      "loss": 0.8597,
+      "step": 158610
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.795679924444377e-05,
+      "loss": 0.9413,
+      "step": 158620
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.795603759501577e-05,
+      "loss": 0.9369,
+      "step": 158630
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.795527594558776e-05,
+      "loss": 0.8496,
+      "step": 158640
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7954514296159766e-05,
+      "loss": 0.8984,
+      "step": 158650
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7953752646731764e-05,
+      "loss": 0.9712,
+      "step": 158660
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.795299099730376e-05,
+      "loss": 0.9444,
+      "step": 158670
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.795222934787576e-05,
+      "loss": 0.9195,
+      "step": 158680
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.795146769844776e-05,
+      "loss": 0.9688,
+      "step": 158690
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.795070604901976e-05,
+      "loss": 0.8269,
+      "step": 158700
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794994439959176e-05,
+      "loss": 0.9285,
+      "step": 158710
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7949182750163755e-05,
+      "loss": 1.013,
+      "step": 158720
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794842110073575e-05,
+      "loss": 0.895,
+      "step": 158730
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794765945130776e-05,
+      "loss": 0.9016,
+      "step": 158740
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7946897801879755e-05,
+      "loss": 0.8937,
+      "step": 158750
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794613615245175e-05,
+      "loss": 0.8331,
+      "step": 158760
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794537450302375e-05,
+      "loss": 0.9369,
+      "step": 158770
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7944612853595754e-05,
+      "loss": 0.8178,
+      "step": 158780
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7943851204167745e-05,
+      "loss": 0.7482,
+      "step": 158790
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794308955473975e-05,
+      "loss": 0.9298,
+      "step": 158800
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7942327905311746e-05,
+      "loss": 0.8248,
+      "step": 158810
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7941566255883744e-05,
+      "loss": 0.887,
+      "step": 158820
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.794080460645574e-05,
+      "loss": 0.9557,
+      "step": 158830
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7940042957027745e-05,
+      "loss": 1.0314,
+      "step": 158840
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7939281307599736e-05,
+      "loss": 0.9394,
+      "step": 158850
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.793851965817174e-05,
+      "loss": 0.8962,
+      "step": 158860
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.793775800874374e-05,
+      "loss": 0.8424,
+      "step": 158870
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7936996359315735e-05,
+      "loss": 0.9757,
+      "step": 158880
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793623470988773e-05,
+      "loss": 0.9478,
+      "step": 158890
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793547306045973e-05,
+      "loss": 1.0292,
+      "step": 158900
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.7934711411031734e-05,
+      "loss": 0.8606,
+      "step": 158910
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793394976160373e-05,
+      "loss": 1.028,
+      "step": 158920
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793318811217573e-05,
+      "loss": 0.8929,
+      "step": 158930
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793242646274773e-05,
+      "loss": 0.9675,
+      "step": 158940
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793166481331973e-05,
+      "loss": 0.8238,
+      "step": 158950
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.793090316389172e-05,
+      "loss": 0.9033,
+      "step": 158960
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.7930141514463726e-05,
+      "loss": 0.9828,
+      "step": 158970
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.7929379865035723e-05,
+      "loss": 1.0383,
+      "step": 158980
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792861821560772e-05,
+      "loss": 0.9152,
+      "step": 158990
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792785656617972e-05,
+      "loss": 0.8329,
+      "step": 159000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792709491675172e-05,
+      "loss": 0.8651,
+      "step": 159010
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792633326732371e-05,
+      "loss": 0.8615,
+      "step": 159020
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792557161789572e-05,
+      "loss": 0.9972,
+      "step": 159030
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.7924809968467715e-05,
+      "loss": 0.8943,
+      "step": 159040
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792404831903971e-05,
+      "loss": 0.9689,
+      "step": 159050
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792328666961171e-05,
+      "loss": 0.9142,
+      "step": 159060
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.7922525020183714e-05,
+      "loss": 0.9785,
+      "step": 159070
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792176337075571e-05,
+      "loss": 0.9549,
+      "step": 159080
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792100172132771e-05,
+      "loss": 0.8946,
+      "step": 159090
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.792024007189971e-05,
+      "loss": 0.8675,
+      "step": 159100
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7919478422471704e-05,
+      "loss": 0.8667,
+      "step": 159110
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.791871677304371e-05,
+      "loss": 0.8734,
+      "step": 159120
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7917955123615706e-05,
+      "loss": 0.9276,
+      "step": 159130
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.79171934741877e-05,
+      "loss": 0.8342,
+      "step": 159140
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.79164318247597e-05,
+      "loss": 0.9229,
+      "step": 159150
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7915670175331705e-05,
+      "loss": 1.0927,
+      "step": 159160
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7914908525903696e-05,
+      "loss": 0.8971,
+      "step": 159170
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.79141468764757e-05,
+      "loss": 0.9284,
+      "step": 159180
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.79133852270477e-05,
+      "loss": 0.8853,
+      "step": 159190
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7912623577619695e-05,
+      "loss": 0.8545,
+      "step": 159200
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.791186192819169e-05,
+      "loss": 0.8865,
+      "step": 159210
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7911100278763697e-05,
+      "loss": 0.8562,
+      "step": 159220
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.791033862933569e-05,
+      "loss": 0.9262,
+      "step": 159230
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.790957697990769e-05,
+      "loss": 0.8792,
+      "step": 159240
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.790881533047969e-05,
+      "loss": 0.8437,
+      "step": 159250
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7908053681051686e-05,
+      "loss": 0.9332,
+      "step": 159260
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7907292031623684e-05,
+      "loss": 0.8152,
+      "step": 159270
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.790653038219569e-05,
+      "loss": 0.9259,
+      "step": 159280
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.7905768732767686e-05,
+      "loss": 0.8994,
+      "step": 159290
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.790500708333968e-05,
+      "loss": 0.8955,
+      "step": 159300
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.790424543391168e-05,
+      "loss": 1.035,
+      "step": 159310
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.790348378448368e-05,
+      "loss": 0.9657,
+      "step": 159320
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.790272213505568e-05,
+      "loss": 0.9168,
+      "step": 159330
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.790196048562767e-05,
+      "loss": 0.7818,
+      "step": 159340
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.790119883619968e-05,
+      "loss": 0.9329,
+      "step": 159350
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7900437186771675e-05,
+      "loss": 0.9536,
+      "step": 159360
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789967553734367e-05,
+      "loss": 0.9042,
+      "step": 159370
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789891388791567e-05,
+      "loss": 0.8675,
+      "step": 159380
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7898152238487674e-05,
+      "loss": 1.032,
+      "step": 159390
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7897390589059665e-05,
+      "loss": 1.0497,
+      "step": 159400
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789662893963167e-05,
+      "loss": 0.8917,
+      "step": 159410
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7895867290203666e-05,
+      "loss": 0.9997,
+      "step": 159420
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7895105640775664e-05,
+      "loss": 0.9802,
+      "step": 159430
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789434399134766e-05,
+      "loss": 1.0915,
+      "step": 159440
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7893582341919665e-05,
+      "loss": 0.8602,
+      "step": 159450
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789282069249166e-05,
+      "loss": 0.9046,
+      "step": 159460
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789205904306366e-05,
+      "loss": 0.9928,
+      "step": 159470
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.789129739363566e-05,
+      "loss": 0.8366,
+      "step": 159480
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7890535744207655e-05,
+      "loss": 0.8296,
+      "step": 159490
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.788977409477966e-05,
+      "loss": 0.9313,
+      "step": 159500
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.788901244535166e-05,
+      "loss": 1.0323,
+      "step": 159510
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.7888250795923654e-05,
+      "loss": 0.9606,
+      "step": 159520
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.788748914649565e-05,
+      "loss": 0.8642,
+      "step": 159530
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7886727497067656e-05,
+      "loss": 0.8291,
+      "step": 159540
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.788596584763965e-05,
+      "loss": 1.0301,
+      "step": 159550
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.788520419821165e-05,
+      "loss": 0.9868,
+      "step": 159560
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.788444254878365e-05,
+      "loss": 0.8758,
+      "step": 159570
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7883680899355646e-05,
+      "loss": 0.8958,
+      "step": 159580
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7882919249927644e-05,
+      "loss": 0.8051,
+      "step": 159590
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.788215760049965e-05,
+      "loss": 0.9037,
+      "step": 159600
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.788139595107164e-05,
+      "loss": 1.0132,
+      "step": 159610
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.788063430164364e-05,
+      "loss": 0.8141,
+      "step": 159620
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787987265221564e-05,
+      "loss": 0.8186,
+      "step": 159630
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787911100278764e-05,
+      "loss": 0.9,
+      "step": 159640
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7878349353359635e-05,
+      "loss": 0.8645,
+      "step": 159650
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787758770393164e-05,
+      "loss": 1.0372,
+      "step": 159660
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787682605450364e-05,
+      "loss": 0.8918,
+      "step": 159670
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7876064405075634e-05,
+      "loss": 0.9197,
+      "step": 159680
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787530275564763e-05,
+      "loss": 0.7962,
+      "step": 159690
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787454110621963e-05,
+      "loss": 0.9438,
+      "step": 159700
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7873779456791633e-05,
+      "loss": 0.906,
+      "step": 159710
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787301780736363e-05,
+      "loss": 0.8511,
+      "step": 159720
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787225615793563e-05,
+      "loss": 1.0113,
+      "step": 159730
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.7871494508507626e-05,
+      "loss": 0.9324,
+      "step": 159740
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.787073285907963e-05,
+      "loss": 0.9154,
+      "step": 159750
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786997120965162e-05,
+      "loss": 0.8532,
+      "step": 159760
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7869209560223625e-05,
+      "loss": 0.9179,
+      "step": 159770
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7868447910795616e-05,
+      "loss": 0.9123,
+      "step": 159780
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786768626136762e-05,
+      "loss": 0.9002,
+      "step": 159790
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786692461193962e-05,
+      "loss": 1.0008,
+      "step": 159800
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7866162962511615e-05,
+      "loss": 0.865,
+      "step": 159810
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786540131308361e-05,
+      "loss": 0.8242,
+      "step": 159820
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7864639663655617e-05,
+      "loss": 0.8938,
+      "step": 159830
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786387801422761e-05,
+      "loss": 0.9041,
+      "step": 159840
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786311636479961e-05,
+      "loss": 0.8709,
+      "step": 159850
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786235471537161e-05,
+      "loss": 0.966,
+      "step": 159860
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7861593065943606e-05,
+      "loss": 0.8723,
+      "step": 159870
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786083141651561e-05,
+      "loss": 0.8934,
+      "step": 159880
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.786006976708761e-05,
+      "loss": 0.8622,
+      "step": 159890
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7859308117659606e-05,
+      "loss": 0.955,
+      "step": 159900
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.78585464682316e-05,
+      "loss": 0.8915,
+      "step": 159910
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.785778481880361e-05,
+      "loss": 0.8852,
+      "step": 159920
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.78570231693756e-05,
+      "loss": 0.7874,
+      "step": 159930
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.78562615199476e-05,
+      "loss": 0.832,
+      "step": 159940
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.78554998705196e-05,
+      "loss": 0.9322,
+      "step": 159950
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.78547382210916e-05,
+      "loss": 0.839,
+      "step": 159960
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7853976571663595e-05,
+      "loss": 0.8678,
+      "step": 159970
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.78532149222356e-05,
+      "loss": 0.8941,
+      "step": 159980
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.785245327280759e-05,
+      "loss": 0.8257,
+      "step": 159990
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7851691623379594e-05,
+      "loss": 0.8255,
+      "step": 160000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.785092997395159e-05,
+      "loss": 0.8869,
+      "step": 160010
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.785016832452359e-05,
+      "loss": 1.0097,
+      "step": 160020
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7849406675095586e-05,
+      "loss": 0.9266,
+      "step": 160030
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784864502566759e-05,
+      "loss": 0.8912,
+      "step": 160040
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784788337623959e-05,
+      "loss": 0.8632,
+      "step": 160050
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7847121726811585e-05,
+      "loss": 0.8487,
+      "step": 160060
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784636007738358e-05,
+      "loss": 0.8942,
+      "step": 160070
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784559842795558e-05,
+      "loss": 0.9305,
+      "step": 160080
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7844836778527585e-05,
+      "loss": 0.8861,
+      "step": 160090
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784407512909958e-05,
+      "loss": 0.9636,
+      "step": 160100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784331347967158e-05,
+      "loss": 1.0263,
+      "step": 160110
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784255183024358e-05,
+      "loss": 0.9211,
+      "step": 160120
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784179018081558e-05,
+      "loss": 0.969,
+      "step": 160130
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.784102853138757e-05,
+      "loss": 0.9141,
+      "step": 160140
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7840266881959576e-05,
+      "loss": 0.8649,
+      "step": 160150
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7839505232531574e-05,
+      "loss": 0.8999,
+      "step": 160160
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.783874358310357e-05,
+      "loss": 0.9721,
+      "step": 160170
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.783798193367557e-05,
+      "loss": 0.8605,
+      "step": 160180
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.783722028424757e-05,
+      "loss": 0.9023,
+      "step": 160190
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7836458634819564e-05,
+      "loss": 0.8509,
+      "step": 160200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783569698539157e-05,
+      "loss": 0.839,
+      "step": 160210
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7834935335963565e-05,
+      "loss": 0.9513,
+      "step": 160220
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783417368653556e-05,
+      "loss": 0.8413,
+      "step": 160230
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783341203710756e-05,
+      "loss": 1.0061,
+      "step": 160240
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783265038767956e-05,
+      "loss": 0.9747,
+      "step": 160250
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783188873825156e-05,
+      "loss": 0.878,
+      "step": 160260
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783112708882356e-05,
+      "loss": 0.9034,
+      "step": 160270
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.783036543939556e-05,
+      "loss": 0.8468,
+      "step": 160280
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7829603789967554e-05,
+      "loss": 1.0275,
+      "step": 160290
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782884214053956e-05,
+      "loss": 0.8972,
+      "step": 160300
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782808049111155e-05,
+      "loss": 0.8681,
+      "step": 160310
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7827318841683553e-05,
+      "loss": 0.9502,
+      "step": 160320
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782655719225555e-05,
+      "loss": 0.9442,
+      "step": 160330
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782579554282755e-05,
+      "loss": 0.8685,
+      "step": 160340
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7825033893399546e-05,
+      "loss": 0.9334,
+      "step": 160350
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782427224397155e-05,
+      "loss": 0.8925,
+      "step": 160360
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782351059454354e-05,
+      "loss": 0.9148,
+      "step": 160370
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7822748945115545e-05,
+      "loss": 1.1029,
+      "step": 160380
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782198729568754e-05,
+      "loss": 0.9366,
+      "step": 160390
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782122564625954e-05,
+      "loss": 0.8329,
+      "step": 160400
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.782046399683154e-05,
+      "loss": 0.8566,
+      "step": 160410
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781970234740354e-05,
+      "loss": 0.8937,
+      "step": 160420
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781894069797553e-05,
+      "loss": 0.831,
+      "step": 160430
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7818179048547537e-05,
+      "loss": 0.8376,
+      "step": 160440
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7817417399119534e-05,
+      "loss": 0.909,
+      "step": 160450
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781665574969153e-05,
+      "loss": 1.0277,
+      "step": 160460
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7815894100263536e-05,
+      "loss": 0.806,
+      "step": 160470
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781513245083553e-05,
+      "loss": 0.9433,
+      "step": 160480
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781437080140753e-05,
+      "loss": 0.8999,
+      "step": 160490
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781360915197953e-05,
+      "loss": 0.9247,
+      "step": 160500
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781284750255153e-05,
+      "loss": 0.8873,
+      "step": 160510
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781208585312352e-05,
+      "loss": 0.8715,
+      "step": 160520
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.781132420369553e-05,
+      "loss": 0.8995,
+      "step": 160530
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7810562554267525e-05,
+      "loss": 0.9765,
+      "step": 160540
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.780980090483952e-05,
+      "loss": 0.926,
+      "step": 160550
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.780903925541152e-05,
+      "loss": 0.97,
+      "step": 160560
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7808277605983524e-05,
+      "loss": 0.8362,
+      "step": 160570
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7807515956555515e-05,
+      "loss": 0.87,
+      "step": 160580
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.780675430712752e-05,
+      "loss": 0.8468,
+      "step": 160590
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7805992657699516e-05,
+      "loss": 1.0036,
+      "step": 160600
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7805231008271514e-05,
+      "loss": 0.9774,
+      "step": 160610
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.780446935884351e-05,
+      "loss": 1.0823,
+      "step": 160620
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.7803707709415516e-05,
+      "loss": 0.9863,
+      "step": 160630
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7802946059987506e-05,
+      "loss": 0.9085,
+      "step": 160640
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.780218441055951e-05,
+      "loss": 0.9252,
+      "step": 160650
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.780142276113151e-05,
+      "loss": 0.9219,
+      "step": 160660
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7800661111703505e-05,
+      "loss": 0.9602,
+      "step": 160670
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.779989946227551e-05,
+      "loss": 0.8799,
+      "step": 160680
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.779913781284751e-05,
+      "loss": 0.8469,
+      "step": 160690
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7798376163419505e-05,
+      "loss": 0.8266,
+      "step": 160700
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.77976145139915e-05,
+      "loss": 0.9176,
+      "step": 160710
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.77968528645635e-05,
+      "loss": 0.8957,
+      "step": 160720
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.77960912151355e-05,
+      "loss": 0.8264,
+      "step": 160730
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.77953295657075e-05,
+      "loss": 0.9004,
+      "step": 160740
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.779456791627949e-05,
+      "loss": 0.9039,
+      "step": 160750
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7793806266851496e-05,
+      "loss": 1.0016,
+      "step": 160760
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7793044617423494e-05,
+      "loss": 0.894,
+      "step": 160770
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.779228296799549e-05,
+      "loss": 0.947,
+      "step": 160780
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.779152131856749e-05,
+      "loss": 0.9197,
+      "step": 160790
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.779075966913949e-05,
+      "loss": 0.8412,
+      "step": 160800
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7789998019711484e-05,
+      "loss": 1.0025,
+      "step": 160810
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.778923637028349e-05,
+      "loss": 0.9771,
+      "step": 160820
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7788474720855485e-05,
+      "loss": 0.9183,
+      "step": 160830
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.778771307142748e-05,
+      "loss": 0.8937,
+      "step": 160840
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.778695142199949e-05,
+      "loss": 0.8927,
+      "step": 160850
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7786189772571484e-05,
+      "loss": 0.9757,
+      "step": 160860
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.778542812314348e-05,
+      "loss": 0.8378,
+      "step": 160870
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.778466647371548e-05,
+      "loss": 0.8916,
+      "step": 160880
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7783904824287484e-05,
+      "loss": 0.8972,
+      "step": 160890
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7783143174859474e-05,
+      "loss": 0.7798,
+      "step": 160900
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.778238152543148e-05,
+      "loss": 0.9274,
+      "step": 160910
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7781619876003476e-05,
+      "loss": 0.9687,
+      "step": 160920
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7780858226575473e-05,
+      "loss": 0.9847,
+      "step": 160930
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.778009657714747e-05,
+      "loss": 0.8162,
+      "step": 160940
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7779334927719475e-05,
+      "loss": 1.0059,
+      "step": 160950
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7778573278291466e-05,
+      "loss": 0.8742,
+      "step": 160960
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777781162886347e-05,
+      "loss": 0.9697,
+      "step": 160970
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777704997943547e-05,
+      "loss": 0.8421,
+      "step": 160980
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7776288330007465e-05,
+      "loss": 0.8749,
+      "step": 160990
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777552668057946e-05,
+      "loss": 0.9373,
+      "step": 161000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777476503115147e-05,
+      "loss": 0.9212,
+      "step": 161010
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777400338172346e-05,
+      "loss": 0.8801,
+      "step": 161020
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777324173229546e-05,
+      "loss": 0.8656,
+      "step": 161030
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777248008286746e-05,
+      "loss": 0.9052,
+      "step": 161040
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.7771718433439457e-05,
+      "loss": 0.8356,
+      "step": 161050
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777095678401146e-05,
+      "loss": 0.8833,
+      "step": 161060
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.777019513458346e-05,
+      "loss": 0.9068,
+      "step": 161070
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7769433485155456e-05,
+      "loss": 0.7641,
+      "step": 161080
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776867183572745e-05,
+      "loss": 0.9186,
+      "step": 161090
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776791018629946e-05,
+      "loss": 0.9063,
+      "step": 161100
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776714853687145e-05,
+      "loss": 0.9789,
+      "step": 161110
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776638688744345e-05,
+      "loss": 0.8968,
+      "step": 161120
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776562523801545e-05,
+      "loss": 0.9007,
+      "step": 161130
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776486358858745e-05,
+      "loss": 0.8889,
+      "step": 161140
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7764101939159445e-05,
+      "loss": 0.9704,
+      "step": 161150
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776334028973144e-05,
+      "loss": 0.9611,
+      "step": 161160
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776257864030344e-05,
+      "loss": 0.7939,
+      "step": 161170
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7761816990875444e-05,
+      "loss": 1.0051,
+      "step": 161180
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7761055341447435e-05,
+      "loss": 0.8525,
+      "step": 161190
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.776029369201944e-05,
+      "loss": 1.0045,
+      "step": 161200
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7759532042591436e-05,
+      "loss": 0.9126,
+      "step": 161210
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7758770393163434e-05,
+      "loss": 0.8902,
+      "step": 161220
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.775800874373543e-05,
+      "loss": 0.9753,
+      "step": 161230
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7757247094307436e-05,
+      "loss": 0.9395,
+      "step": 161240
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.775648544487943e-05,
+      "loss": 0.9841,
+      "step": 161250
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.775572379545143e-05,
+      "loss": 0.8163,
+      "step": 161260
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7754962146023435e-05,
+      "loss": 0.8665,
+      "step": 161270
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.7754200496595425e-05,
+      "loss": 0.8816,
+      "step": 161280
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.775343884716743e-05,
+      "loss": 0.8764,
+      "step": 161290
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.775267719773943e-05,
+      "loss": 0.9949,
+      "step": 161300
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7751915548311425e-05,
+      "loss": 0.8947,
+      "step": 161310
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.775115389888342e-05,
+      "loss": 0.8344,
+      "step": 161320
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7750392249455426e-05,
+      "loss": 0.9744,
+      "step": 161330
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774963060002742e-05,
+      "loss": 0.967,
+      "step": 161340
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774886895059942e-05,
+      "loss": 0.9418,
+      "step": 161350
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774810730117142e-05,
+      "loss": 0.9402,
+      "step": 161360
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7747345651743416e-05,
+      "loss": 0.8502,
+      "step": 161370
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7746584002315414e-05,
+      "loss": 0.8761,
+      "step": 161380
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774582235288742e-05,
+      "loss": 0.8771,
+      "step": 161390
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774506070345941e-05,
+      "loss": 0.8889,
+      "step": 161400
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774429905403141e-05,
+      "loss": 0.8263,
+      "step": 161410
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774353740460341e-05,
+      "loss": 0.9522,
+      "step": 161420
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774277575517541e-05,
+      "loss": 0.9749,
+      "step": 161430
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7742014105747405e-05,
+      "loss": 0.8818,
+      "step": 161440
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774125245631941e-05,
+      "loss": 0.8931,
+      "step": 161450
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.774049080689141e-05,
+      "loss": 0.8889,
+      "step": 161460
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7739729157463404e-05,
+      "loss": 0.8366,
+      "step": 161470
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.773896750803541e-05,
+      "loss": 0.8165,
+      "step": 161480
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.77382058586074e-05,
+      "loss": 0.956,
+      "step": 161490
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.7737444209179404e-05,
+      "loss": 0.9115,
+      "step": 161500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.77366825597514e-05,
+      "loss": 0.8754,
+      "step": 161510
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.77359209103234e-05,
+      "loss": 0.9867,
+      "step": 161520
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.7735159260895396e-05,
+      "loss": 0.8938,
+      "step": 161530
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.77343976114674e-05,
+      "loss": 0.9986,
+      "step": 161540
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.773363596203939e-05,
+      "loss": 0.8876,
+      "step": 161550
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.7732874312611395e-05,
+      "loss": 0.8569,
+      "step": 161560
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.773211266318339e-05,
+      "loss": 0.9308,
+      "step": 161570
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.773135101375539e-05,
+      "loss": 0.983,
+      "step": 161580
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.773058936432739e-05,
+      "loss": 0.9572,
+      "step": 161590
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772982771489939e-05,
+      "loss": 0.9051,
+      "step": 161600
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772906606547138e-05,
+      "loss": 0.8811,
+      "step": 161610
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772830441604339e-05,
+      "loss": 0.9278,
+      "step": 161620
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.7727542766615384e-05,
+      "loss": 1.0429,
+      "step": 161630
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772678111718738e-05,
+      "loss": 0.8908,
+      "step": 161640
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.7726019467759386e-05,
+      "loss": 0.9633,
+      "step": 161650
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772525781833138e-05,
+      "loss": 0.8382,
+      "step": 161660
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772449616890338e-05,
+      "loss": 0.8016,
+      "step": 161670
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772373451947538e-05,
+      "loss": 0.9269,
+      "step": 161680
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.7722972870047376e-05,
+      "loss": 0.9157,
+      "step": 161690
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772221122061937e-05,
+      "loss": 0.9608,
+      "step": 161700
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772144957119138e-05,
+      "loss": 0.8548,
+      "step": 161710
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.772068792176337e-05,
+      "loss": 0.9443,
+      "step": 161720
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771992627233537e-05,
+      "loss": 0.9385,
+      "step": 161730
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771916462290737e-05,
+      "loss": 0.8525,
+      "step": 161740
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771840297347937e-05,
+      "loss": 1.0648,
+      "step": 161750
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.7717641324051365e-05,
+      "loss": 1.042,
+      "step": 161760
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771687967462337e-05,
+      "loss": 0.8638,
+      "step": 161770
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771611802519536e-05,
+      "loss": 1.022,
+      "step": 161780
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.7715356375767364e-05,
+      "loss": 0.9544,
+      "step": 161790
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771459472633936e-05,
+      "loss": 0.9257,
+      "step": 161800
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771383307691136e-05,
+      "loss": 0.9626,
+      "step": 161810
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.7713071427483356e-05,
+      "loss": 0.8797,
+      "step": 161820
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771230977805536e-05,
+      "loss": 0.8922,
+      "step": 161830
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771154812862736e-05,
+      "loss": 1.062,
+      "step": 161840
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.7710786479199356e-05,
+      "loss": 0.8696,
+      "step": 161850
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.771002482977136e-05,
+      "loss": 0.7707,
+      "step": 161860
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.770926318034335e-05,
+      "loss": 0.899,
+      "step": 161870
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.7708501530915355e-05,
+      "loss": 0.9741,
+      "step": 161880
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.770773988148735e-05,
+      "loss": 0.8647,
+      "step": 161890
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.770697823205935e-05,
+      "loss": 0.903,
+      "step": 161900
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.770621658263135e-05,
+      "loss": 0.9996,
+      "step": 161910
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.770545493320335e-05,
+      "loss": 1.0442,
+      "step": 161920
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.770469328377534e-05,
+      "loss": 0.886,
+      "step": 161930
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.7703931634347346e-05,
+      "loss": 0.9285,
+      "step": 161940
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7703169984919344e-05,
+      "loss": 1.0183,
+      "step": 161950
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.770240833549134e-05,
+      "loss": 0.8675,
+      "step": 161960
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.770164668606334e-05,
+      "loss": 0.983,
+      "step": 161970
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.770088503663534e-05,
+      "loss": 0.8194,
+      "step": 161980
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7700123387207334e-05,
+      "loss": 0.8578,
+      "step": 161990
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769936173777934e-05,
+      "loss": 0.8942,
+      "step": 162000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7698600088351335e-05,
+      "loss": 0.9032,
+      "step": 162010
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769783843892333e-05,
+      "loss": 0.9249,
+      "step": 162020
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769707678949533e-05,
+      "loss": 0.8741,
+      "step": 162030
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7696315140067335e-05,
+      "loss": 0.9193,
+      "step": 162040
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769555349063933e-05,
+      "loss": 0.8914,
+      "step": 162050
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769479184121133e-05,
+      "loss": 0.8674,
+      "step": 162060
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7694030191783334e-05,
+      "loss": 0.8867,
+      "step": 162070
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7693268542355324e-05,
+      "loss": 0.9534,
+      "step": 162080
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769250689292733e-05,
+      "loss": 1.0585,
+      "step": 162090
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769174524349932e-05,
+      "loss": 0.8604,
+      "step": 162100
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7690983594071324e-05,
+      "loss": 1.1203,
+      "step": 162110
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.769022194464332e-05,
+      "loss": 0.8505,
+      "step": 162120
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.768946029521532e-05,
+      "loss": 0.8616,
+      "step": 162130
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7688698645787316e-05,
+      "loss": 0.9791,
+      "step": 162140
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.768793699635932e-05,
+      "loss": 0.8766,
+      "step": 162150
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.768717534693131e-05,
+      "loss": 0.8822,
+      "step": 162160
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7686413697503315e-05,
+      "loss": 0.9355,
+      "step": 162170
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.768565204807531e-05,
+      "loss": 0.9714,
+      "step": 162180
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.768489039864731e-05,
+      "loss": 0.8875,
+      "step": 162190
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.768412874921931e-05,
+      "loss": 0.8783,
+      "step": 162200
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.768336709979131e-05,
+      "loss": 0.8435,
+      "step": 162210
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.768260545036331e-05,
+      "loss": 0.9251,
+      "step": 162220
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.768184380093531e-05,
+      "loss": 0.87,
+      "step": 162230
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7681082151507304e-05,
+      "loss": 0.9086,
+      "step": 162240
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.76803205020793e-05,
+      "loss": 0.8507,
+      "step": 162250
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7679558852651306e-05,
+      "loss": 0.8644,
+      "step": 162260
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7678797203223303e-05,
+      "loss": 0.9386,
+      "step": 162270
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.76780355537953e-05,
+      "loss": 0.953,
+      "step": 162280
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.76772739043673e-05,
+      "loss": 0.9656,
+      "step": 162290
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.76765122549393e-05,
+      "loss": 0.9695,
+      "step": 162300
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.767575060551129e-05,
+      "loss": 0.8919,
+      "step": 162310
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.76749889560833e-05,
+      "loss": 0.9768,
+      "step": 162320
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7674227306655295e-05,
+      "loss": 0.9133,
+      "step": 162330
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.767346565722729e-05,
+      "loss": 1.0086,
+      "step": 162340
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.767270400779929e-05,
+      "loss": 1.0122,
+      "step": 162350
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7671942358371294e-05,
+      "loss": 0.8854,
+      "step": 162360
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7671180708943285e-05,
+      "loss": 0.9284,
+      "step": 162370
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.767041905951529e-05,
+      "loss": 0.8573,
+      "step": 162380
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7669657410087287e-05,
+      "loss": 0.9218,
+      "step": 162390
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7668895760659284e-05,
+      "loss": 0.7739,
+      "step": 162400
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.766813411123128e-05,
+      "loss": 0.8375,
+      "step": 162410
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7667372461803286e-05,
+      "loss": 0.9573,
+      "step": 162420
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.766661081237528e-05,
+      "loss": 0.9216,
+      "step": 162430
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.766584916294728e-05,
+      "loss": 0.9372,
+      "step": 162440
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7665087513519285e-05,
+      "loss": 0.9453,
+      "step": 162450
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7664325864091276e-05,
+      "loss": 0.8339,
+      "step": 162460
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.766356421466328e-05,
+      "loss": 0.9276,
+      "step": 162470
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.766280256523528e-05,
+      "loss": 0.9188,
+      "step": 162480
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7662040915807275e-05,
+      "loss": 0.8505,
+      "step": 162490
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.766127926637927e-05,
+      "loss": 1.0231,
+      "step": 162500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7660517616951276e-05,
+      "loss": 0.8361,
+      "step": 162510
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765975596752327e-05,
+      "loss": 0.9908,
+      "step": 162520
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765899431809527e-05,
+      "loss": 0.9037,
+      "step": 162530
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765823266866727e-05,
+      "loss": 0.9097,
+      "step": 162540
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7657471019239266e-05,
+      "loss": 0.845,
+      "step": 162550
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.7656709369811264e-05,
+      "loss": 0.9068,
+      "step": 162560
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765594772038326e-05,
+      "loss": 0.8798,
+      "step": 162570
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765518607095526e-05,
+      "loss": 0.8771,
+      "step": 162580
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765442442152726e-05,
+      "loss": 0.8124,
+      "step": 162590
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.765366277209926e-05,
+      "loss": 0.8828,
+      "step": 162600
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.765290112267126e-05,
+      "loss": 0.8795,
+      "step": 162610
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7652139473243255e-05,
+      "loss": 0.8605,
+      "step": 162620
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.765137782381525e-05,
+      "loss": 0.9072,
+      "step": 162630
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.765061617438726e-05,
+      "loss": 0.907,
+      "step": 162640
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7649854524959255e-05,
+      "loss": 0.8623,
+      "step": 162650
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.764909287553125e-05,
+      "loss": 0.976,
+      "step": 162660
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.764833122610325e-05,
+      "loss": 0.9902,
+      "step": 162670
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7647569576675254e-05,
+      "loss": 0.9634,
+      "step": 162680
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7646807927247244e-05,
+      "loss": 1.0179,
+      "step": 162690
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.764604627781925e-05,
+      "loss": 0.8595,
+      "step": 162700
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7645284628391246e-05,
+      "loss": 0.9373,
+      "step": 162710
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7644522978963244e-05,
+      "loss": 0.9315,
+      "step": 162720
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.764376132953524e-05,
+      "loss": 0.8434,
+      "step": 162730
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7642999680107245e-05,
+      "loss": 0.9799,
+      "step": 162740
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7642238030679236e-05,
+      "loss": 0.7833,
+      "step": 162750
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.764147638125124e-05,
+      "loss": 0.892,
+      "step": 162760
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.764071473182324e-05,
+      "loss": 1.0346,
+      "step": 162770
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7639953082395235e-05,
+      "loss": 0.8274,
+      "step": 162780
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.763919143296723e-05,
+      "loss": 0.9057,
+      "step": 162790
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.763842978353924e-05,
+      "loss": 0.945,
+      "step": 162800
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.7637668134111234e-05,
+      "loss": 0.8643,
+      "step": 162810
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.763690648468323e-05,
+      "loss": 0.926,
+      "step": 162820
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763614483525523e-05,
+      "loss": 0.8992,
+      "step": 162830
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763538318582723e-05,
+      "loss": 0.9206,
+      "step": 162840
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763462153639923e-05,
+      "loss": 0.9089,
+      "step": 162850
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763385988697123e-05,
+      "loss": 0.8451,
+      "step": 162860
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7633098237543226e-05,
+      "loss": 0.853,
+      "step": 162870
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7632336588115223e-05,
+      "loss": 0.9085,
+      "step": 162880
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763157493868723e-05,
+      "loss": 0.8597,
+      "step": 162890
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763081328925922e-05,
+      "loss": 0.8529,
+      "step": 162900
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.763005163983122e-05,
+      "loss": 0.9146,
+      "step": 162910
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762928999040322e-05,
+      "loss": 0.9004,
+      "step": 162920
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762852834097522e-05,
+      "loss": 0.9317,
+      "step": 162930
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7627766691547215e-05,
+      "loss": 1.0088,
+      "step": 162940
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762700504211922e-05,
+      "loss": 0.9091,
+      "step": 162950
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762624339269121e-05,
+      "loss": 1.0209,
+      "step": 162960
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7625481743263214e-05,
+      "loss": 0.9035,
+      "step": 162970
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762472009383521e-05,
+      "loss": 0.9741,
+      "step": 162980
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762395844440721e-05,
+      "loss": 0.9069,
+      "step": 162990
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7623196794979207e-05,
+      "loss": 0.916,
+      "step": 163000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7622435145551204e-05,
+      "loss": 0.9232,
+      "step": 163010
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.762167349612321e-05,
+      "loss": 0.9516,
+      "step": 163020
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.7620911846695206e-05,
+      "loss": 0.8963,
+      "step": 163030
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.76201501972672e-05,
+      "loss": 0.8492,
+      "step": 163040
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.76193885478392e-05,
+      "loss": 0.848,
+      "step": 163050
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7618626898411205e-05,
+      "loss": 0.9602,
+      "step": 163060
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7617865248983196e-05,
+      "loss": 0.8654,
+      "step": 163070
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.76171035995552e-05,
+      "loss": 0.8742,
+      "step": 163080
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.76163419501272e-05,
+      "loss": 0.8762,
+      "step": 163090
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7615580300699195e-05,
+      "loss": 0.8139,
+      "step": 163100
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.761481865127119e-05,
+      "loss": 0.8984,
+      "step": 163110
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7614057001843196e-05,
+      "loss": 0.9357,
+      "step": 163120
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.761329535241519e-05,
+      "loss": 1.0458,
+      "step": 163130
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.761253370298719e-05,
+      "loss": 0.8677,
+      "step": 163140
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.761177205355919e-05,
+      "loss": 0.8768,
+      "step": 163150
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7611010404131186e-05,
+      "loss": 0.9094,
+      "step": 163160
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7610248754703184e-05,
+      "loss": 0.8614,
+      "step": 163170
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760948710527519e-05,
+      "loss": 1.0431,
+      "step": 163180
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7608725455847186e-05,
+      "loss": 0.8751,
+      "step": 163190
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760796380641918e-05,
+      "loss": 0.8411,
+      "step": 163200
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760720215699118e-05,
+      "loss": 0.8898,
+      "step": 163210
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760644050756318e-05,
+      "loss": 0.9407,
+      "step": 163220
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760567885813518e-05,
+      "loss": 0.8931,
+      "step": 163230
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760491720870718e-05,
+      "loss": 0.8235,
+      "step": 163240
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.760415555927918e-05,
+      "loss": 0.9396,
+      "step": 163250
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7603393909851175e-05,
+      "loss": 0.8639,
+      "step": 163260
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.760263226042318e-05,
+      "loss": 0.8693,
+      "step": 163270
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.760187061099517e-05,
+      "loss": 0.8606,
+      "step": 163280
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7601108961567174e-05,
+      "loss": 0.915,
+      "step": 163290
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.760034731213917e-05,
+      "loss": 0.8986,
+      "step": 163300
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759958566271117e-05,
+      "loss": 0.9007,
+      "step": 163310
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7598824013283166e-05,
+      "loss": 0.9008,
+      "step": 163320
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759806236385517e-05,
+      "loss": 0.8973,
+      "step": 163330
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759730071442716e-05,
+      "loss": 0.857,
+      "step": 163340
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7596539064999165e-05,
+      "loss": 0.9483,
+      "step": 163350
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759577741557116e-05,
+      "loss": 0.9503,
+      "step": 163360
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759501576614316e-05,
+      "loss": 0.9013,
+      "step": 163370
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759425411671516e-05,
+      "loss": 0.8988,
+      "step": 163380
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759349246728716e-05,
+      "loss": 0.9579,
+      "step": 163390
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759273081785916e-05,
+      "loss": 0.9667,
+      "step": 163400
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759196916843116e-05,
+      "loss": 0.9676,
+      "step": 163410
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7591207519003154e-05,
+      "loss": 0.9619,
+      "step": 163420
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.759044586957515e-05,
+      "loss": 0.8638,
+      "step": 163430
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7589684220147156e-05,
+      "loss": 0.841,
+      "step": 163440
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7588922570719154e-05,
+      "loss": 0.8314,
+      "step": 163450
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.758816092129115e-05,
+      "loss": 1.0027,
+      "step": 163460
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.758739927186315e-05,
+      "loss": 0.8795,
+      "step": 163470
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7586637622435146e-05,
+      "loss": 0.9905,
+      "step": 163480
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.7585875973007143e-05,
+      "loss": 0.9155,
+      "step": 163490
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.758511432357915e-05,
+      "loss": 0.8196,
+      "step": 163500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.758435267415114e-05,
+      "loss": 1.0127,
+      "step": 163510
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.758359102472314e-05,
+      "loss": 0.9317,
+      "step": 163520
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.758282937529514e-05,
+      "loss": 0.9619,
+      "step": 163530
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.758206772586714e-05,
+      "loss": 0.9617,
+      "step": 163540
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.7581306076439135e-05,
+      "loss": 0.9328,
+      "step": 163550
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.758054442701114e-05,
+      "loss": 0.9862,
+      "step": 163560
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757978277758314e-05,
+      "loss": 0.9361,
+      "step": 163570
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.7579021128155134e-05,
+      "loss": 0.9745,
+      "step": 163580
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757825947872713e-05,
+      "loss": 0.9189,
+      "step": 163590
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757749782929913e-05,
+      "loss": 0.9613,
+      "step": 163600
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757673617987113e-05,
+      "loss": 0.7413,
+      "step": 163610
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757597453044313e-05,
+      "loss": 0.9594,
+      "step": 163620
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757521288101513e-05,
+      "loss": 0.9497,
+      "step": 163630
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.7574451231587126e-05,
+      "loss": 0.9578,
+      "step": 163640
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757368958215913e-05,
+      "loss": 0.9146,
+      "step": 163650
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757292793273112e-05,
+      "loss": 0.9362,
+      "step": 163660
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.7572166283303125e-05,
+      "loss": 0.9059,
+      "step": 163670
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757140463387512e-05,
+      "loss": 0.9057,
+      "step": 163680
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.757064298444712e-05,
+      "loss": 0.8194,
+      "step": 163690
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.756988133501912e-05,
+      "loss": 0.9087,
+      "step": 163700
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756911968559112e-05,
+      "loss": 0.8597,
+      "step": 163710
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756835803616311e-05,
+      "loss": 0.9277,
+      "step": 163720
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7567596386735117e-05,
+      "loss": 0.9405,
+      "step": 163730
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7566834737307114e-05,
+      "loss": 0.8839,
+      "step": 163740
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756607308787911e-05,
+      "loss": 0.8498,
+      "step": 163750
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756531143845111e-05,
+      "loss": 0.8849,
+      "step": 163760
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756454978902311e-05,
+      "loss": 1.1056,
+      "step": 163770
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756378813959511e-05,
+      "loss": 0.8832,
+      "step": 163780
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756302649016711e-05,
+      "loss": 0.9785,
+      "step": 163790
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7562264840739106e-05,
+      "loss": 0.9143,
+      "step": 163800
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.75615031913111e-05,
+      "loss": 0.8935,
+      "step": 163810
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.756074154188311e-05,
+      "loss": 1.0393,
+      "step": 163820
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7559979892455105e-05,
+      "loss": 0.9042,
+      "step": 163830
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.75592182430271e-05,
+      "loss": 0.8759,
+      "step": 163840
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.75584565935991e-05,
+      "loss": 1.0001,
+      "step": 163850
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7557694944171104e-05,
+      "loss": 0.8655,
+      "step": 163860
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7556933294743095e-05,
+      "loss": 0.9047,
+      "step": 163870
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.75561716453151e-05,
+      "loss": 0.8908,
+      "step": 163880
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7555409995887096e-05,
+      "loss": 0.9095,
+      "step": 163890
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7554648346459094e-05,
+      "loss": 0.9587,
+      "step": 163900
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.755388669703109e-05,
+      "loss": 0.8389,
+      "step": 163910
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7553125047603095e-05,
+      "loss": 0.7978,
+      "step": 163920
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7552363398175086e-05,
+      "loss": 0.8501,
+      "step": 163930
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.755160174874709e-05,
+      "loss": 0.8846,
+      "step": 163940
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.755084009931909e-05,
+      "loss": 0.8482,
+      "step": 163950
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7550078449891085e-05,
+      "loss": 0.9551,
+      "step": 163960
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754931680046308e-05,
+      "loss": 0.8619,
+      "step": 163970
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754855515103508e-05,
+      "loss": 0.8522,
+      "step": 163980
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7547793501607085e-05,
+      "loss": 0.8923,
+      "step": 163990
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754703185217908e-05,
+      "loss": 0.8835,
+      "step": 164000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754627020275108e-05,
+      "loss": 0.9535,
+      "step": 164010
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754550855332308e-05,
+      "loss": 0.9921,
+      "step": 164020
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754474690389508e-05,
+      "loss": 0.9155,
+      "step": 164030
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754398525446707e-05,
+      "loss": 0.8799,
+      "step": 164040
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7543223605039076e-05,
+      "loss": 0.9563,
+      "step": 164050
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7542461955611074e-05,
+      "loss": 0.9128,
+      "step": 164060
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754170030618307e-05,
+      "loss": 0.9668,
+      "step": 164070
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754093865675507e-05,
+      "loss": 0.957,
+      "step": 164080
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.754017700732707e-05,
+      "loss": 0.904,
+      "step": 164090
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7539415357899063e-05,
+      "loss": 0.9912,
+      "step": 164100
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.753865370847107e-05,
+      "loss": 0.8781,
+      "step": 164110
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7537892059043065e-05,
+      "loss": 0.9559,
+      "step": 164120
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.753713040961506e-05,
+      "loss": 1.0276,
+      "step": 164130
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.753636876018706e-05,
+      "loss": 0.8137,
+      "step": 164140
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7535607110759064e-05,
+      "loss": 0.9793,
+      "step": 164150
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.753484546133106e-05,
+      "loss": 0.8665,
+      "step": 164160
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.753408381190306e-05,
+      "loss": 0.8751,
+      "step": 164170
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.753332216247506e-05,
+      "loss": 0.9145,
+      "step": 164180
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7532560513047054e-05,
+      "loss": 0.9279,
+      "step": 164190
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.753179886361906e-05,
+      "loss": 0.8893,
+      "step": 164200
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7531037214191056e-05,
+      "loss": 0.8452,
+      "step": 164210
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.753027556476305e-05,
+      "loss": 0.9229,
+      "step": 164220
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752951391533505e-05,
+      "loss": 0.8651,
+      "step": 164230
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7528752265907055e-05,
+      "loss": 0.8582,
+      "step": 164240
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7527990616479046e-05,
+      "loss": 0.9676,
+      "step": 164250
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752722896705105e-05,
+      "loss": 0.8692,
+      "step": 164260
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752646731762305e-05,
+      "loss": 0.8433,
+      "step": 164270
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7525705668195045e-05,
+      "loss": 0.9601,
+      "step": 164280
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752494401876704e-05,
+      "loss": 0.9818,
+      "step": 164290
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752418236933905e-05,
+      "loss": 0.9318,
+      "step": 164300
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752342071991104e-05,
+      "loss": 0.9352,
+      "step": 164310
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752265907048304e-05,
+      "loss": 0.9815,
+      "step": 164320
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.752189742105504e-05,
+      "loss": 0.9026,
+      "step": 164330
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7521135771627037e-05,
+      "loss": 0.9499,
+      "step": 164340
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.7520374122199034e-05,
+      "loss": 0.9179,
+      "step": 164350
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751961247277104e-05,
+      "loss": 0.942,
+      "step": 164360
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7518850823343036e-05,
+      "loss": 0.8703,
+      "step": 164370
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751808917391503e-05,
+      "loss": 0.8997,
+      "step": 164380
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751732752448703e-05,
+      "loss": 0.8421,
+      "step": 164390
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751656587505903e-05,
+      "loss": 1.0408,
+      "step": 164400
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751580422563103e-05,
+      "loss": 0.9188,
+      "step": 164410
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751504257620302e-05,
+      "loss": 0.8792,
+      "step": 164420
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751428092677503e-05,
+      "loss": 0.8833,
+      "step": 164430
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7513519277347025e-05,
+      "loss": 0.9655,
+      "step": 164440
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751275762791902e-05,
+      "loss": 0.9511,
+      "step": 164450
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.751199597849102e-05,
+      "loss": 0.8472,
+      "step": 164460
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7511234329063024e-05,
+      "loss": 0.8482,
+      "step": 164470
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7510472679635015e-05,
+      "loss": 1.0188,
+      "step": 164480
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.750971103020702e-05,
+      "loss": 0.8722,
+      "step": 164490
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7508949380779016e-05,
+      "loss": 0.829,
+      "step": 164500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7508187731351014e-05,
+      "loss": 0.9198,
+      "step": 164510
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.750742608192301e-05,
+      "loss": 0.9956,
+      "step": 164520
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7506664432495015e-05,
+      "loss": 0.928,
+      "step": 164530
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7505902783067006e-05,
+      "loss": 0.9316,
+      "step": 164540
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.750514113363901e-05,
+      "loss": 0.8622,
+      "step": 164550
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.750437948421101e-05,
+      "loss": 0.9387,
+      "step": 164560
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.7503617834783005e-05,
+      "loss": 0.9059,
+      "step": 164570
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.750285618535501e-05,
+      "loss": 0.9097,
+      "step": 164580
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.750209453592701e-05,
+      "loss": 0.9003,
+      "step": 164590
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7501332886499005e-05,
+      "loss": 0.9533,
+      "step": 164600
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7500571237071e-05,
+      "loss": 0.8655,
+      "step": 164610
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7499809587643006e-05,
+      "loss": 0.9576,
+      "step": 164620
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7499047938215e-05,
+      "loss": 0.8045,
+      "step": 164630
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7498286288787e-05,
+      "loss": 0.9543,
+      "step": 164640
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7497524639359e-05,
+      "loss": 0.9865,
+      "step": 164650
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7496762989930996e-05,
+      "loss": 0.8965,
+      "step": 164660
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7496001340502994e-05,
+      "loss": 0.8691,
+      "step": 164670
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7495239691075e-05,
+      "loss": 1.0251,
+      "step": 164680
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.749447804164699e-05,
+      "loss": 0.9088,
+      "step": 164690
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.749371639221899e-05,
+      "loss": 0.8867,
+      "step": 164700
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.749295474279099e-05,
+      "loss": 0.9203,
+      "step": 164710
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.749219309336299e-05,
+      "loss": 0.9983,
+      "step": 164720
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7491431443934985e-05,
+      "loss": 0.8408,
+      "step": 164730
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.749066979450699e-05,
+      "loss": 0.9467,
+      "step": 164740
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.748990814507899e-05,
+      "loss": 0.9714,
+      "step": 164750
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7489146495650984e-05,
+      "loss": 0.9306,
+      "step": 164760
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.748838484622298e-05,
+      "loss": 0.8418,
+      "step": 164770
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.748762319679498e-05,
+      "loss": 0.9234,
+      "step": 164780
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.7486861547366984e-05,
+      "loss": 0.8458,
+      "step": 164790
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.748609989793898e-05,
+      "loss": 0.9788,
+      "step": 164800
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.748533824851098e-05,
+      "loss": 0.8112,
+      "step": 164810
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.7484576599082976e-05,
+      "loss": 0.872,
+      "step": 164820
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.748381494965498e-05,
+      "loss": 0.9756,
+      "step": 164830
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.748305330022697e-05,
+      "loss": 0.9565,
+      "step": 164840
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.7482291650798975e-05,
+      "loss": 0.916,
+      "step": 164850
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.7481530001370966e-05,
+      "loss": 0.9539,
+      "step": 164860
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.748076835194297e-05,
+      "loss": 0.9793,
+      "step": 164870
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.748000670251497e-05,
+      "loss": 0.9046,
+      "step": 164880
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.7479245053086965e-05,
+      "loss": 0.8742,
+      "step": 164890
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747848340365896e-05,
+      "loss": 1.03,
+      "step": 164900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747772175423097e-05,
+      "loss": 0.8726,
+      "step": 164910
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747696010480296e-05,
+      "loss": 0.9354,
+      "step": 164920
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747619845537496e-05,
+      "loss": 0.8705,
+      "step": 164930
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747543680594696e-05,
+      "loss": 0.8914,
+      "step": 164940
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.7474675156518957e-05,
+      "loss": 0.9544,
+      "step": 164950
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747391350709096e-05,
+      "loss": 0.9929,
+      "step": 164960
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747315185766296e-05,
+      "loss": 1.0202,
+      "step": 164970
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.7472390208234956e-05,
+      "loss": 0.8623,
+      "step": 164980
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747162855880695e-05,
+      "loss": 0.9086,
+      "step": 164990
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747086690937896e-05,
+      "loss": 0.9151,
+      "step": 165000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.747010525995095e-05,
+      "loss": 0.9402,
+      "step": 165010
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746934361052295e-05,
+      "loss": 0.9198,
+      "step": 165020
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746858196109495e-05,
+      "loss": 0.9258,
+      "step": 165030
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746782031166695e-05,
+      "loss": 0.9385,
+      "step": 165040
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.7467058662238945e-05,
+      "loss": 0.9366,
+      "step": 165050
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746629701281095e-05,
+      "loss": 1.0347,
+      "step": 165060
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746553536338294e-05,
+      "loss": 0.9695,
+      "step": 165070
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.7464773713954944e-05,
+      "loss": 0.8597,
+      "step": 165080
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746401206452694e-05,
+      "loss": 0.8984,
+      "step": 165090
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746325041509894e-05,
+      "loss": 0.914,
+      "step": 165100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.7462488765670936e-05,
+      "loss": 0.9035,
+      "step": 165110
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746172711624294e-05,
+      "loss": 0.9435,
+      "step": 165120
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.746096546681493e-05,
+      "loss": 0.8731,
+      "step": 165130
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.7460203817386936e-05,
+      "loss": 0.946,
+      "step": 165140
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745944216795893e-05,
+      "loss": 0.8675,
+      "step": 165150
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745868051853093e-05,
+      "loss": 0.8653,
+      "step": 165160
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.7457918869102935e-05,
+      "loss": 0.9488,
+      "step": 165170
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745715721967493e-05,
+      "loss": 0.9929,
+      "step": 165180
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745639557024693e-05,
+      "loss": 0.9687,
+      "step": 165190
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745563392081893e-05,
+      "loss": 0.9375,
+      "step": 165200
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745487227139093e-05,
+      "loss": 0.973,
+      "step": 165210
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.745411062196292e-05,
+      "loss": 0.9012,
+      "step": 165220
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.7453348972534926e-05,
+      "loss": 0.8518,
+      "step": 165230
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7452587323106924e-05,
+      "loss": 1.0681,
+      "step": 165240
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.745182567367892e-05,
+      "loss": 0.8496,
+      "step": 165250
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.745106402425092e-05,
+      "loss": 0.9751,
+      "step": 165260
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.745030237482292e-05,
+      "loss": 0.8576,
+      "step": 165270
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7449540725394914e-05,
+      "loss": 0.9235,
+      "step": 165280
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744877907596692e-05,
+      "loss": 0.8978,
+      "step": 165290
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7448017426538915e-05,
+      "loss": 0.9027,
+      "step": 165300
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744725577711091e-05,
+      "loss": 0.8715,
+      "step": 165310
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744649412768291e-05,
+      "loss": 0.9638,
+      "step": 165320
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744573247825491e-05,
+      "loss": 0.8112,
+      "step": 165330
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7444970828826905e-05,
+      "loss": 0.9072,
+      "step": 165340
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744420917939891e-05,
+      "loss": 1.0127,
+      "step": 165350
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744344752997091e-05,
+      "loss": 0.8981,
+      "step": 165360
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7442685880542904e-05,
+      "loss": 0.94,
+      "step": 165370
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.744192423111491e-05,
+      "loss": 0.9053,
+      "step": 165380
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.74411625816869e-05,
+      "loss": 1.0157,
+      "step": 165390
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7440400932258904e-05,
+      "loss": 0.8975,
+      "step": 165400
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.74396392828309e-05,
+      "loss": 0.935,
+      "step": 165410
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.74388776334029e-05,
+      "loss": 1.074,
+      "step": 165420
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7438115983974896e-05,
+      "loss": 1.0438,
+      "step": 165430
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.74373543345469e-05,
+      "loss": 1.02,
+      "step": 165440
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.743659268511889e-05,
+      "loss": 0.9629,
+      "step": 165450
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.7435831035690895e-05,
+      "loss": 0.9014,
+      "step": 165460
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.743506938626289e-05,
+      "loss": 0.9962,
+      "step": 165470
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.743430773683489e-05,
+      "loss": 0.9377,
+      "step": 165480
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.743354608740689e-05,
+      "loss": 0.9577,
+      "step": 165490
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.743278443797889e-05,
+      "loss": 0.8972,
+      "step": 165500
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.743202278855088e-05,
+      "loss": 0.9275,
+      "step": 165510
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.743126113912289e-05,
+      "loss": 0.9458,
+      "step": 165520
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.7430499489694884e-05,
+      "loss": 0.8392,
+      "step": 165530
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742973784026688e-05,
+      "loss": 0.9037,
+      "step": 165540
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.7428976190838886e-05,
+      "loss": 0.8262,
+      "step": 165550
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742821454141088e-05,
+      "loss": 0.9257,
+      "step": 165560
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742745289198288e-05,
+      "loss": 0.9382,
+      "step": 165570
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742669124255488e-05,
+      "loss": 1.0295,
+      "step": 165580
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742592959312688e-05,
+      "loss": 0.7692,
+      "step": 165590
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742516794369887e-05,
+      "loss": 0.8513,
+      "step": 165600
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742440629427088e-05,
+      "loss": 0.8482,
+      "step": 165610
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.7423644644842875e-05,
+      "loss": 0.8878,
+      "step": 165620
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742288299541487e-05,
+      "loss": 0.8733,
+      "step": 165630
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.742212134598687e-05,
+      "loss": 0.876,
+      "step": 165640
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.7421359696558874e-05,
+      "loss": 0.939,
+      "step": 165650
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.7420598047130865e-05,
+      "loss": 0.929,
+      "step": 165660
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.741983639770287e-05,
+      "loss": 0.8714,
+      "step": 165670
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7419074748274866e-05,
+      "loss": 0.9214,
+      "step": 165680
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7418313098846864e-05,
+      "loss": 0.9369,
+      "step": 165690
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.741755144941886e-05,
+      "loss": 0.8747,
+      "step": 165700
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7416789799990866e-05,
+      "loss": 0.9668,
+      "step": 165710
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7416028150562856e-05,
+      "loss": 1.0079,
+      "step": 165720
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.741526650113486e-05,
+      "loss": 0.9479,
+      "step": 165730
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.741450485170686e-05,
+      "loss": 0.9743,
+      "step": 165740
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7413743202278856e-05,
+      "loss": 0.898,
+      "step": 165750
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.741298155285086e-05,
+      "loss": 0.8918,
+      "step": 165760
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.741221990342286e-05,
+      "loss": 0.8985,
+      "step": 165770
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7411458253994855e-05,
+      "loss": 0.9697,
+      "step": 165780
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.741069660456685e-05,
+      "loss": 0.98,
+      "step": 165790
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740993495513885e-05,
+      "loss": 0.9884,
+      "step": 165800
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740917330571085e-05,
+      "loss": 0.9364,
+      "step": 165810
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740841165628285e-05,
+      "loss": 0.8827,
+      "step": 165820
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740765000685484e-05,
+      "loss": 0.923,
+      "step": 165830
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7406888357426846e-05,
+      "loss": 1.0308,
+      "step": 165840
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7406126707998844e-05,
+      "loss": 0.9401,
+      "step": 165850
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740536505857084e-05,
+      "loss": 0.9252,
+      "step": 165860
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740460340914284e-05,
+      "loss": 0.8419,
+      "step": 165870
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.740384175971484e-05,
+      "loss": 0.8738,
+      "step": 165880
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.7403080110286834e-05,
+      "loss": 0.8272,
+      "step": 165890
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.740231846085884e-05,
+      "loss": 0.9019,
+      "step": 165900
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7401556811430835e-05,
+      "loss": 0.9243,
+      "step": 165910
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.740079516200283e-05,
+      "loss": 0.89,
+      "step": 165920
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.740003351257483e-05,
+      "loss": 1.1129,
+      "step": 165930
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7399271863146834e-05,
+      "loss": 0.8791,
+      "step": 165940
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.739851021371883e-05,
+      "loss": 0.9227,
+      "step": 165950
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.739774856429083e-05,
+      "loss": 0.8786,
+      "step": 165960
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7396986914862834e-05,
+      "loss": 1.0327,
+      "step": 165970
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7396225265434824e-05,
+      "loss": 0.7949,
+      "step": 165980
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.739546361600683e-05,
+      "loss": 0.8216,
+      "step": 165990
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7394701966578826e-05,
+      "loss": 0.8814,
+      "step": 166000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7393940317150824e-05,
+      "loss": 0.9111,
+      "step": 166010
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.739317866772282e-05,
+      "loss": 0.8685,
+      "step": 166020
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7392417018294825e-05,
+      "loss": 0.8221,
+      "step": 166030
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7391655368866816e-05,
+      "loss": 0.915,
+      "step": 166040
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.739089371943882e-05,
+      "loss": 0.9449,
+      "step": 166050
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.739013207001082e-05,
+      "loss": 0.8648,
+      "step": 166060
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.7389370420582815e-05,
+      "loss": 1.0104,
+      "step": 166070
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.738860877115481e-05,
+      "loss": 0.9435,
+      "step": 166080
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.738784712172682e-05,
+      "loss": 0.9715,
+      "step": 166090
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.738708547229881e-05,
+      "loss": 0.9837,
+      "step": 166100
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.738632382287081e-05,
+      "loss": 0.8169,
+      "step": 166110
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.738556217344281e-05,
+      "loss": 0.9045,
+      "step": 166120
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.738480052401481e-05,
+      "loss": 0.8266,
+      "step": 166130
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7384038874586804e-05,
+      "loss": 0.9231,
+      "step": 166140
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.738327722515881e-05,
+      "loss": 0.928,
+      "step": 166150
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7382515575730806e-05,
+      "loss": 0.9565,
+      "step": 166160
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.73817539263028e-05,
+      "loss": 0.9537,
+      "step": 166170
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.738099227687481e-05,
+      "loss": 0.9464,
+      "step": 166180
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.73802306274468e-05,
+      "loss": 0.9263,
+      "step": 166190
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.73794689780188e-05,
+      "loss": 0.877,
+      "step": 166200
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.73787073285908e-05,
+      "loss": 0.8968,
+      "step": 166210
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.73779456791628e-05,
+      "loss": 0.8628,
+      "step": 166220
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7377184029734795e-05,
+      "loss": 0.9125,
+      "step": 166230
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.737642238030679e-05,
+      "loss": 0.8825,
+      "step": 166240
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.737566073087879e-05,
+      "loss": 0.8389,
+      "step": 166250
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7374899081450794e-05,
+      "loss": 0.9889,
+      "step": 166260
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7374137432022785e-05,
+      "loss": 0.9224,
+      "step": 166270
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.737337578259479e-05,
+      "loss": 0.8691,
+      "step": 166280
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7372614133166786e-05,
+      "loss": 0.8261,
+      "step": 166290
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7371852483738784e-05,
+      "loss": 0.8022,
+      "step": 166300
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.737109083431078e-05,
+      "loss": 0.9788,
+      "step": 166310
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7370329184882786e-05,
+      "loss": 0.8385,
+      "step": 166320
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736956753545478e-05,
+      "loss": 0.9187,
+      "step": 166330
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736880588602678e-05,
+      "loss": 0.8587,
+      "step": 166340
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7368044236598785e-05,
+      "loss": 0.8395,
+      "step": 166350
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7367282587170776e-05,
+      "loss": 0.9108,
+      "step": 166360
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736652093774278e-05,
+      "loss": 0.8945,
+      "step": 166370
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736575928831478e-05,
+      "loss": 0.8655,
+      "step": 166380
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7364997638886775e-05,
+      "loss": 0.8119,
+      "step": 166390
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736423598945877e-05,
+      "loss": 0.9635,
+      "step": 166400
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7363474340030776e-05,
+      "loss": 0.9675,
+      "step": 166410
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736271269060277e-05,
+      "loss": 0.906,
+      "step": 166420
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736195104117477e-05,
+      "loss": 0.944,
+      "step": 166430
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.736118939174677e-05,
+      "loss": 0.9851,
+      "step": 166440
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7360427742318766e-05,
+      "loss": 0.8547,
+      "step": 166450
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7359666092890764e-05,
+      "loss": 0.8549,
+      "step": 166460
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735890444346277e-05,
+      "loss": 0.9095,
+      "step": 166470
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735814279403476e-05,
+      "loss": 0.9434,
+      "step": 166480
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735738114460676e-05,
+      "loss": 0.9884,
+      "step": 166490
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735661949517876e-05,
+      "loss": 0.953,
+      "step": 166500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735585784575076e-05,
+      "loss": 0.9655,
+      "step": 166510
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.7355096196322755e-05,
+      "loss": 0.9245,
+      "step": 166520
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735433454689476e-05,
+      "loss": 0.8842,
+      "step": 166530
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.735357289746676e-05,
+      "loss": 0.9436,
+      "step": 166540
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.7352811248038754e-05,
+      "loss": 0.9205,
+      "step": 166550
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.735204959861076e-05,
+      "loss": 0.9298,
+      "step": 166560
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.735128794918275e-05,
+      "loss": 0.9029,
+      "step": 166570
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.7350526299754754e-05,
+      "loss": 0.8568,
+      "step": 166580
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734976465032675e-05,
+      "loss": 1.141,
+      "step": 166590
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734900300089875e-05,
+      "loss": 0.9249,
+      "step": 166600
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.7348241351470746e-05,
+      "loss": 0.9114,
+      "step": 166610
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734747970204275e-05,
+      "loss": 0.8428,
+      "step": 166620
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734671805261474e-05,
+      "loss": 0.8042,
+      "step": 166630
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.7345956403186745e-05,
+      "loss": 0.8911,
+      "step": 166640
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734519475375874e-05,
+      "loss": 0.9026,
+      "step": 166650
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734443310433074e-05,
+      "loss": 0.7671,
+      "step": 166660
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734367145490274e-05,
+      "loss": 0.8382,
+      "step": 166670
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734290980547474e-05,
+      "loss": 0.9746,
+      "step": 166680
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734214815604673e-05,
+      "loss": 0.9318,
+      "step": 166690
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.734138650661874e-05,
+      "loss": 0.8411,
+      "step": 166700
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.7340624857190734e-05,
+      "loss": 0.9644,
+      "step": 166710
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.733986320776273e-05,
+      "loss": 0.9376,
+      "step": 166720
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.733910155833473e-05,
+      "loss": 0.9413,
+      "step": 166730
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.733833990890673e-05,
+      "loss": 0.8868,
+      "step": 166740
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.733757825947873e-05,
+      "loss": 0.9431,
+      "step": 166750
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.733681661005073e-05,
+      "loss": 0.8672,
+      "step": 166760
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7336054960622726e-05,
+      "loss": 0.8815,
+      "step": 166770
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.733529331119472e-05,
+      "loss": 0.8373,
+      "step": 166780
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.733453166176673e-05,
+      "loss": 0.9226,
+      "step": 166790
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.733377001233872e-05,
+      "loss": 1.0006,
+      "step": 166800
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.733300836291072e-05,
+      "loss": 0.8782,
+      "step": 166810
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.733224671348272e-05,
+      "loss": 0.8442,
+      "step": 166820
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.733148506405472e-05,
+      "loss": 0.9144,
+      "step": 166830
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7330723414626715e-05,
+      "loss": 0.8979,
+      "step": 166840
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732996176519872e-05,
+      "loss": 0.9747,
+      "step": 166850
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732920011577071e-05,
+      "loss": 0.9669,
+      "step": 166860
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7328438466342714e-05,
+      "loss": 0.8282,
+      "step": 166870
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732767681691471e-05,
+      "loss": 0.8702,
+      "step": 166880
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732691516748671e-05,
+      "loss": 1.0214,
+      "step": 166890
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7326153518058706e-05,
+      "loss": 0.887,
+      "step": 166900
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732539186863071e-05,
+      "loss": 0.9625,
+      "step": 166910
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732463021920271e-05,
+      "loss": 0.94,
+      "step": 166920
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7323868569774706e-05,
+      "loss": 0.8176,
+      "step": 166930
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.732310692034671e-05,
+      "loss": 0.8641,
+      "step": 166940
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.73223452709187e-05,
+      "loss": 0.938,
+      "step": 166950
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7321583621490705e-05,
+      "loss": 0.9087,
+      "step": 166960
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.73208219720627e-05,
+      "loss": 0.9774,
+      "step": 166970
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.73200603226347e-05,
+      "loss": 0.8692,
+      "step": 166980
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.73192986732067e-05,
+      "loss": 0.871,
+      "step": 166990
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.73185370237787e-05,
+      "loss": 0.8364,
+      "step": 167000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731777537435069e-05,
+      "loss": 0.8751,
+      "step": 167010
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7317013724922696e-05,
+      "loss": 0.8244,
+      "step": 167020
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7316252075494694e-05,
+      "loss": 0.9453,
+      "step": 167030
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731549042606669e-05,
+      "loss": 1.0368,
+      "step": 167040
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731472877663869e-05,
+      "loss": 0.916,
+      "step": 167050
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731396712721069e-05,
+      "loss": 0.8459,
+      "step": 167060
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7313205477782684e-05,
+      "loss": 0.9981,
+      "step": 167070
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731244382835469e-05,
+      "loss": 0.9415,
+      "step": 167080
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7311682178926685e-05,
+      "loss": 0.8435,
+      "step": 167090
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731092052949868e-05,
+      "loss": 0.897,
+      "step": 167100
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.731015888007068e-05,
+      "loss": 0.9739,
+      "step": 167110
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7309397230642685e-05,
+      "loss": 0.9282,
+      "step": 167120
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.730863558121468e-05,
+      "loss": 0.8359,
+      "step": 167130
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.730787393178668e-05,
+      "loss": 0.9378,
+      "step": 167140
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7307112282358684e-05,
+      "loss": 0.9589,
+      "step": 167150
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7306350632930675e-05,
+      "loss": 0.9012,
+      "step": 167160
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.730558898350268e-05,
+      "loss": 0.8867,
+      "step": 167170
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.730482733407467e-05,
+      "loss": 0.9073,
+      "step": 167180
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7304065684646674e-05,
+      "loss": 0.8888,
+      "step": 167190
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.730330403521867e-05,
+      "loss": 0.8818,
+      "step": 167200
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.730254238579067e-05,
+      "loss": 0.8624,
+      "step": 167210
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.7301780736362666e-05,
+      "loss": 0.9179,
+      "step": 167220
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.730101908693467e-05,
+      "loss": 1.056,
+      "step": 167230
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.730025743750666e-05,
+      "loss": 0.9146,
+      "step": 167240
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.7299495788078665e-05,
+      "loss": 0.9111,
+      "step": 167250
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729873413865066e-05,
+      "loss": 0.9747,
+      "step": 167260
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729797248922266e-05,
+      "loss": 0.8328,
+      "step": 167270
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729721083979466e-05,
+      "loss": 0.8401,
+      "step": 167280
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729644919036666e-05,
+      "loss": 0.8748,
+      "step": 167290
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729568754093866e-05,
+      "loss": 1.0431,
+      "step": 167300
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729492589151066e-05,
+      "loss": 0.9438,
+      "step": 167310
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.7294164242082654e-05,
+      "loss": 0.977,
+      "step": 167320
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729340259265465e-05,
+      "loss": 0.9904,
+      "step": 167330
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.7292640943226656e-05,
+      "loss": 0.8898,
+      "step": 167340
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.7291879293798653e-05,
+      "loss": 0.9334,
+      "step": 167350
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729111764437065e-05,
+      "loss": 0.9309,
+      "step": 167360
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.729035599494265e-05,
+      "loss": 0.7647,
+      "step": 167370
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.728959434551465e-05,
+      "loss": 0.9012,
+      "step": 167380
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.728883269608664e-05,
+      "loss": 0.9561,
+      "step": 167390
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.728807104665865e-05,
+      "loss": 0.871,
+      "step": 167400
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.7287309397230645e-05,
+      "loss": 0.8859,
+      "step": 167410
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.728654774780264e-05,
+      "loss": 0.8856,
+      "step": 167420
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.728578609837464e-05,
+      "loss": 0.8248,
+      "step": 167430
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7285024448946644e-05,
+      "loss": 0.8612,
+      "step": 167440
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7284262799518635e-05,
+      "loss": 0.9153,
+      "step": 167450
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.728350115009064e-05,
+      "loss": 0.8474,
+      "step": 167460
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.728273950066264e-05,
+      "loss": 0.9609,
+      "step": 167470
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7281977851234634e-05,
+      "loss": 0.9176,
+      "step": 167480
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.728121620180663e-05,
+      "loss": 0.8707,
+      "step": 167490
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7280454552378636e-05,
+      "loss": 0.9545,
+      "step": 167500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727969290295063e-05,
+      "loss": 0.9509,
+      "step": 167510
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727893125352263e-05,
+      "loss": 1.0289,
+      "step": 167520
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727816960409463e-05,
+      "loss": 1.0454,
+      "step": 167530
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7277407954666626e-05,
+      "loss": 0.926,
+      "step": 167540
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727664630523863e-05,
+      "loss": 0.9847,
+      "step": 167550
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727588465581063e-05,
+      "loss": 0.9422,
+      "step": 167560
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7275123006382625e-05,
+      "loss": 0.9339,
+      "step": 167570
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727436135695462e-05,
+      "loss": 0.9653,
+      "step": 167580
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7273599707526627e-05,
+      "loss": 0.8534,
+      "step": 167590
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727283805809862e-05,
+      "loss": 1.0241,
+      "step": 167600
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727207640867062e-05,
+      "loss": 0.9717,
+      "step": 167610
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.727131475924261e-05,
+      "loss": 0.8962,
+      "step": 167620
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7270553109814616e-05,
+      "loss": 0.934,
+      "step": 167630
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.7269791460386614e-05,
+      "loss": 0.9732,
+      "step": 167640
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.726902981095861e-05,
+      "loss": 0.814,
+      "step": 167650
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.726826816153061e-05,
+      "loss": 0.8849,
+      "step": 167660
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.726750651210261e-05,
+      "loss": 0.9367,
+      "step": 167670
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.726674486267461e-05,
+      "loss": 0.8528,
+      "step": 167680
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.726598321324661e-05,
+      "loss": 0.8462,
+      "step": 167690
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7265221563818605e-05,
+      "loss": 0.9961,
+      "step": 167700
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.72644599143906e-05,
+      "loss": 0.8998,
+      "step": 167710
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.726369826496261e-05,
+      "loss": 0.881,
+      "step": 167720
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7262936615534605e-05,
+      "loss": 0.8159,
+      "step": 167730
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.72621749661066e-05,
+      "loss": 0.9182,
+      "step": 167740
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.72614133166786e-05,
+      "loss": 1.0157,
+      "step": 167750
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7260651667250604e-05,
+      "loss": 0.9427,
+      "step": 167760
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7259890017822595e-05,
+      "loss": 0.8604,
+      "step": 167770
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.72591283683946e-05,
+      "loss": 0.8321,
+      "step": 167780
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7258366718966596e-05,
+      "loss": 0.931,
+      "step": 167790
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7257605069538594e-05,
+      "loss": 0.8208,
+      "step": 167800
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.725684342011059e-05,
+      "loss": 0.938,
+      "step": 167810
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7256081770682595e-05,
+      "loss": 1.041,
+      "step": 167820
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7255320121254586e-05,
+      "loss": 0.954,
+      "step": 167830
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.725455847182659e-05,
+      "loss": 0.8611,
+      "step": 167840
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.725379682239859e-05,
+      "loss": 0.8843,
+      "step": 167850
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7253035172970585e-05,
+      "loss": 0.9458,
+      "step": 167860
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.725227352354258e-05,
+      "loss": 0.9516,
+      "step": 167870
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.725151187411459e-05,
+      "loss": 1.0004,
+      "step": 167880
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.7250750224686584e-05,
+      "loss": 0.9037,
+      "step": 167890
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724998857525858e-05,
+      "loss": 0.8138,
+      "step": 167900
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724922692583058e-05,
+      "loss": 0.8426,
+      "step": 167910
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724846527640258e-05,
+      "loss": 0.7299,
+      "step": 167920
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724770362697458e-05,
+      "loss": 0.8808,
+      "step": 167930
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724694197754658e-05,
+      "loss": 0.8869,
+      "step": 167940
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.7246180328118576e-05,
+      "loss": 0.9079,
+      "step": 167950
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.7245418678690573e-05,
+      "loss": 0.8939,
+      "step": 167960
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724465702926258e-05,
+      "loss": 0.8777,
+      "step": 167970
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724389537983457e-05,
+      "loss": 0.8529,
+      "step": 167980
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724313373040657e-05,
+      "loss": 0.9732,
+      "step": 167990
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724237208097857e-05,
+      "loss": 0.9456,
+      "step": 168000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724161043155057e-05,
+      "loss": 0.8585,
+      "step": 168010
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.7240848782122565e-05,
+      "loss": 0.9633,
+      "step": 168020
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.724008713269457e-05,
+      "loss": 0.8484,
+      "step": 168030
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.723932548326656e-05,
+      "loss": 0.9061,
+      "step": 168040
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.7238563833838564e-05,
+      "loss": 0.9041,
+      "step": 168050
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.723780218441056e-05,
+      "loss": 0.8461,
+      "step": 168060
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.723704053498256e-05,
+      "loss": 0.8747,
+      "step": 168070
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.723627888555456e-05,
+      "loss": 0.8171,
+      "step": 168080
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7235517236126554e-05,
+      "loss": 0.9874,
+      "step": 168090
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.723475558669856e-05,
+      "loss": 0.8445,
+      "step": 168100
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7233993937270556e-05,
+      "loss": 0.8731,
+      "step": 168110
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.723323228784255e-05,
+      "loss": 0.927,
+      "step": 168120
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.723247063841455e-05,
+      "loss": 0.9161,
+      "step": 168130
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7231708988986555e-05,
+      "loss": 0.9813,
+      "step": 168140
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7230947339558546e-05,
+      "loss": 1.0064,
+      "step": 168150
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.723018569013055e-05,
+      "loss": 0.9086,
+      "step": 168160
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722942404070255e-05,
+      "loss": 0.9265,
+      "step": 168170
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7228662391274545e-05,
+      "loss": 0.9333,
+      "step": 168180
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722790074184654e-05,
+      "loss": 0.881,
+      "step": 168190
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7227139092418547e-05,
+      "loss": 0.8563,
+      "step": 168200
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722637744299054e-05,
+      "loss": 0.8869,
+      "step": 168210
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722561579356254e-05,
+      "loss": 0.8991,
+      "step": 168220
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722485414413454e-05,
+      "loss": 0.8976,
+      "step": 168230
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7224092494706536e-05,
+      "loss": 0.8724,
+      "step": 168240
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7223330845278534e-05,
+      "loss": 0.8355,
+      "step": 168250
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722256919585054e-05,
+      "loss": 0.8733,
+      "step": 168260
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7221807546422536e-05,
+      "loss": 0.9543,
+      "step": 168270
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722104589699453e-05,
+      "loss": 0.8763,
+      "step": 168280
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.722028424756653e-05,
+      "loss": 0.8774,
+      "step": 168290
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721952259813853e-05,
+      "loss": 0.8718,
+      "step": 168300
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721876094871053e-05,
+      "loss": 0.8442,
+      "step": 168310
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721799929928253e-05,
+      "loss": 0.9158,
+      "step": 168320
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721723764985453e-05,
+      "loss": 0.7964,
+      "step": 168330
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.7216476000426525e-05,
+      "loss": 0.8257,
+      "step": 168340
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721571435099853e-05,
+      "loss": 0.8584,
+      "step": 168350
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721495270157052e-05,
+      "loss": 0.8332,
+      "step": 168360
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.7214191052142524e-05,
+      "loss": 0.9013,
+      "step": 168370
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721342940271452e-05,
+      "loss": 0.8419,
+      "step": 168380
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721266775328652e-05,
+      "loss": 0.8634,
+      "step": 168390
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.7211906103858516e-05,
+      "loss": 0.9805,
+      "step": 168400
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721114445443052e-05,
+      "loss": 0.9189,
+      "step": 168410
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.721038280500251e-05,
+      "loss": 1.0178,
+      "step": 168420
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.7209621155574515e-05,
+      "loss": 0.7768,
+      "step": 168430
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.720885950614651e-05,
+      "loss": 0.9151,
+      "step": 168440
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.720809785671851e-05,
+      "loss": 0.9401,
+      "step": 168450
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.720733620729051e-05,
+      "loss": 0.963,
+      "step": 168460
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.720657455786251e-05,
+      "loss": 0.956,
+      "step": 168470
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.720581290843451e-05,
+      "loss": 0.8794,
+      "step": 168480
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.720505125900651e-05,
+      "loss": 0.9381,
+      "step": 168490
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.7204289609578504e-05,
+      "loss": 0.8664,
+      "step": 168500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.72035279601505e-05,
+      "loss": 0.8329,
+      "step": 168510
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7202766310722506e-05,
+      "loss": 0.9043,
+      "step": 168520
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7202004661294504e-05,
+      "loss": 0.8918,
+      "step": 168530
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.72012430118665e-05,
+      "loss": 0.9904,
+      "step": 168540
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.72004813624385e-05,
+      "loss": 0.8053,
+      "step": 168550
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7199719713010496e-05,
+      "loss": 0.9296,
+      "step": 168560
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7198958063582494e-05,
+      "loss": 0.994,
+      "step": 168570
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.71981964141545e-05,
+      "loss": 0.9796,
+      "step": 168580
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719743476472649e-05,
+      "loss": 0.8952,
+      "step": 168590
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719667311529849e-05,
+      "loss": 0.9049,
+      "step": 168600
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719591146587049e-05,
+      "loss": 0.9268,
+      "step": 168610
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719514981644249e-05,
+      "loss": 0.8754,
+      "step": 168620
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7194388167014485e-05,
+      "loss": 1.0124,
+      "step": 168630
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719362651758649e-05,
+      "loss": 0.9405,
+      "step": 168640
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719286486815849e-05,
+      "loss": 0.8319,
+      "step": 168650
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7192103218730484e-05,
+      "loss": 0.8767,
+      "step": 168660
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719134156930248e-05,
+      "loss": 0.9187,
+      "step": 168670
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.719057991987448e-05,
+      "loss": 0.8668,
+      "step": 168680
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7189818270446483e-05,
+      "loss": 0.7723,
+      "step": 168690
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.718905662101848e-05,
+      "loss": 0.8218,
+      "step": 168700
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.718829497159048e-05,
+      "loss": 0.9055,
+      "step": 168710
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7187533322162476e-05,
+      "loss": 0.9485,
+      "step": 168720
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.718677167273448e-05,
+      "loss": 0.9613,
+      "step": 168730
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718601002330647e-05,
+      "loss": 1.0132,
+      "step": 168740
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7185248373878475e-05,
+      "loss": 0.9402,
+      "step": 168750
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718448672445047e-05,
+      "loss": 0.8386,
+      "step": 168760
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718372507502247e-05,
+      "loss": 0.9313,
+      "step": 168770
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718296342559447e-05,
+      "loss": 0.8866,
+      "step": 168780
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718220177616647e-05,
+      "loss": 0.9879,
+      "step": 168790
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718144012673846e-05,
+      "loss": 0.9762,
+      "step": 168800
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7180678477310467e-05,
+      "loss": 0.8595,
+      "step": 168810
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7179916827882464e-05,
+      "loss": 1.0139,
+      "step": 168820
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717915517845446e-05,
+      "loss": 0.9408,
+      "step": 168830
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717839352902646e-05,
+      "loss": 1.002,
+      "step": 168840
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717763187959846e-05,
+      "loss": 0.9857,
+      "step": 168850
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717687023017046e-05,
+      "loss": 0.8281,
+      "step": 168860
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717610858074246e-05,
+      "loss": 0.8573,
+      "step": 168870
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7175346931314456e-05,
+      "loss": 1.0295,
+      "step": 168880
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717458528188645e-05,
+      "loss": 0.8279,
+      "step": 168890
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717382363245846e-05,
+      "loss": 0.9191,
+      "step": 168900
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7173061983030455e-05,
+      "loss": 0.8827,
+      "step": 168910
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717230033360245e-05,
+      "loss": 0.8439,
+      "step": 168920
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.717153868417445e-05,
+      "loss": 0.8363,
+      "step": 168930
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7170777034746454e-05,
+      "loss": 0.9302,
+      "step": 168940
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7170015385318445e-05,
+      "loss": 0.8963,
+      "step": 168950
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716925373589045e-05,
+      "loss": 1.0072,
+      "step": 168960
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7168492086462446e-05,
+      "loss": 0.9465,
+      "step": 168970
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7167730437034444e-05,
+      "loss": 0.898,
+      "step": 168980
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716696878760644e-05,
+      "loss": 0.8968,
+      "step": 168990
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716620713817844e-05,
+      "loss": 0.8684,
+      "step": 169000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7165445488750436e-05,
+      "loss": 0.9991,
+      "step": 169010
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716468383932244e-05,
+      "loss": 0.898,
+      "step": 169020
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716392218989443e-05,
+      "loss": 0.9038,
+      "step": 169030
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7163160540466435e-05,
+      "loss": 0.9024,
+      "step": 169040
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716239889103843e-05,
+      "loss": 0.8499,
+      "step": 169050
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716163724161043e-05,
+      "loss": 1.0902,
+      "step": 169060
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7160875592182435e-05,
+      "loss": 0.84,
+      "step": 169070
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.716011394275443e-05,
+      "loss": 0.9957,
+      "step": 169080
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715935229332643e-05,
+      "loss": 0.8775,
+      "step": 169090
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715859064389843e-05,
+      "loss": 0.977,
+      "step": 169100
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715782899447043e-05,
+      "loss": 0.9704,
+      "step": 169110
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715706734504242e-05,
+      "loss": 0.925,
+      "step": 169120
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7156305695614426e-05,
+      "loss": 0.8539,
+      "step": 169130
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7155544046186424e-05,
+      "loss": 0.9467,
+      "step": 169140
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715478239675842e-05,
+      "loss": 0.9296,
+      "step": 169150
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715402074733042e-05,
+      "loss": 0.8127,
+      "step": 169160
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.715325909790242e-05,
+      "loss": 0.9255,
+      "step": 169170
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7152497448474414e-05,
+      "loss": 0.9993,
+      "step": 169180
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.715173579904642e-05,
+      "loss": 0.9197,
+      "step": 169190
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7150974149618415e-05,
+      "loss": 0.861,
+      "step": 169200
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.715021250019041e-05,
+      "loss": 0.9071,
+      "step": 169210
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.714945085076241e-05,
+      "loss": 0.9245,
+      "step": 169220
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7148689201334414e-05,
+      "loss": 0.9531,
+      "step": 169230
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.714792755190641e-05,
+      "loss": 0.8875,
+      "step": 169240
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.714716590247841e-05,
+      "loss": 1.009,
+      "step": 169250
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.714640425305041e-05,
+      "loss": 0.9448,
+      "step": 169260
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7145642603622404e-05,
+      "loss": 0.9404,
+      "step": 169270
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.714488095419441e-05,
+      "loss": 0.8519,
+      "step": 169280
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7144119304766406e-05,
+      "loss": 0.8401,
+      "step": 169290
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7143357655338403e-05,
+      "loss": 0.9748,
+      "step": 169300
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.71425960059104e-05,
+      "loss": 0.9649,
+      "step": 169310
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7141834356482405e-05,
+      "loss": 0.8739,
+      "step": 169320
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7141072707054396e-05,
+      "loss": 0.9306,
+      "step": 169330
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.71403110576264e-05,
+      "loss": 0.9722,
+      "step": 169340
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.71395494081984e-05,
+      "loss": 0.9391,
+      "step": 169350
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.7138787758770395e-05,
+      "loss": 0.9055,
+      "step": 169360
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.713802610934239e-05,
+      "loss": 0.9122,
+      "step": 169370
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.71372644599144e-05,
+      "loss": 0.8578,
+      "step": 169380
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.713650281048639e-05,
+      "loss": 0.9219,
+      "step": 169390
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.713574116105839e-05,
+      "loss": 0.8972,
+      "step": 169400
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.713497951163039e-05,
+      "loss": 0.9863,
+      "step": 169410
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7134217862202387e-05,
+      "loss": 0.9638,
+      "step": 169420
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7133456212774384e-05,
+      "loss": 0.9177,
+      "step": 169430
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.713269456334639e-05,
+      "loss": 0.8423,
+      "step": 169440
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7131932913918386e-05,
+      "loss": 0.9071,
+      "step": 169450
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.713117126449038e-05,
+      "loss": 0.8159,
+      "step": 169460
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.713040961506238e-05,
+      "loss": 0.8998,
+      "step": 169470
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712964796563438e-05,
+      "loss": 0.9203,
+      "step": 169480
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712888631620638e-05,
+      "loss": 0.8245,
+      "step": 169490
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712812466677837e-05,
+      "loss": 0.8761,
+      "step": 169500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712736301735038e-05,
+      "loss": 0.9216,
+      "step": 169510
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7126601367922375e-05,
+      "loss": 0.8801,
+      "step": 169520
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712583971849437e-05,
+      "loss": 0.8827,
+      "step": 169530
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712507806906637e-05,
+      "loss": 0.9666,
+      "step": 169540
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7124316419638374e-05,
+      "loss": 0.9415,
+      "step": 169550
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7123554770210365e-05,
+      "loss": 0.8527,
+      "step": 169560
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712279312078237e-05,
+      "loss": 0.9184,
+      "step": 169570
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7122031471354366e-05,
+      "loss": 0.9127,
+      "step": 169580
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7121269821926364e-05,
+      "loss": 0.9014,
+      "step": 169590
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.712050817249836e-05,
+      "loss": 0.9124,
+      "step": 169600
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7119746523070366e-05,
+      "loss": 0.8591,
+      "step": 169610
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7118984873642356e-05,
+      "loss": 0.8383,
+      "step": 169620
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711822322421436e-05,
+      "loss": 0.977,
+      "step": 169630
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711746157478636e-05,
+      "loss": 0.9336,
+      "step": 169640
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7116699925358355e-05,
+      "loss": 0.9345,
+      "step": 169650
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711593827593036e-05,
+      "loss": 0.935,
+      "step": 169660
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711517662650236e-05,
+      "loss": 1.0993,
+      "step": 169670
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7114414977074355e-05,
+      "loss": 0.9144,
+      "step": 169680
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711365332764635e-05,
+      "loss": 0.9247,
+      "step": 169690
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7112891678218356e-05,
+      "loss": 0.907,
+      "step": 169700
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711213002879035e-05,
+      "loss": 0.9257,
+      "step": 169710
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711136837936235e-05,
+      "loss": 0.9524,
+      "step": 169720
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.711060672993435e-05,
+      "loss": 0.8897,
+      "step": 169730
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7109845080506346e-05,
+      "loss": 0.8688,
+      "step": 169740
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7109083431078344e-05,
+      "loss": 0.8844,
+      "step": 169750
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710832178165035e-05,
+      "loss": 0.8593,
+      "step": 169760
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710756013222234e-05,
+      "loss": 0.9963,
+      "step": 169770
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710679848279434e-05,
+      "loss": 0.8486,
+      "step": 169780
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710603683336634e-05,
+      "loss": 0.811,
+      "step": 169790
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710527518393834e-05,
+      "loss": 0.9152,
+      "step": 169800
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7104513534510335e-05,
+      "loss": 0.8447,
+      "step": 169810
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710375188508234e-05,
+      "loss": 0.8317,
+      "step": 169820
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.710299023565433e-05,
+      "loss": 0.9407,
+      "step": 169830
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7102228586226334e-05,
+      "loss": 0.9036,
+      "step": 169840
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.710146693679833e-05,
+      "loss": 0.9482,
+      "step": 169850
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.710070528737033e-05,
+      "loss": 0.9362,
+      "step": 169860
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7099943637942334e-05,
+      "loss": 0.9345,
+      "step": 169870
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709918198851433e-05,
+      "loss": 0.9709,
+      "step": 169880
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709842033908633e-05,
+      "loss": 0.9451,
+      "step": 169890
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7097658689658326e-05,
+      "loss": 0.798,
+      "step": 169900
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709689704023033e-05,
+      "loss": 0.9568,
+      "step": 169910
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709613539080232e-05,
+      "loss": 0.9427,
+      "step": 169920
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7095373741374325e-05,
+      "loss": 0.9518,
+      "step": 169930
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7094612091946316e-05,
+      "loss": 0.9021,
+      "step": 169940
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709385044251832e-05,
+      "loss": 0.9127,
+      "step": 169950
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709308879309032e-05,
+      "loss": 0.9257,
+      "step": 169960
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7092327143662315e-05,
+      "loss": 1.0349,
+      "step": 169970
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709156549423431e-05,
+      "loss": 0.9188,
+      "step": 169980
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709080384480632e-05,
+      "loss": 0.8903,
+      "step": 169990
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.709004219537831e-05,
+      "loss": 1.0547,
+      "step": 170000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.708928054595031e-05,
+      "loss": 0.9653,
+      "step": 170010
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.708851889652231e-05,
+      "loss": 0.8638,
+      "step": 170020
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7087757247094307e-05,
+      "loss": 0.8804,
+      "step": 170030
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.708699559766631e-05,
+      "loss": 0.9594,
+      "step": 170040
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.708623394823831e-05,
+      "loss": 0.9392,
+      "step": 170050
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.7085472298810306e-05,
+      "loss": 0.8937,
+      "step": 170060
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.70847106493823e-05,
+      "loss": 1.02,
+      "step": 170070
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.708394899995431e-05,
+      "loss": 0.8673,
+      "step": 170080
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.70831873505263e-05,
+      "loss": 0.8909,
+      "step": 170090
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.70824257010983e-05,
+      "loss": 0.8853,
+      "step": 170100
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.70816640516703e-05,
+      "loss": 0.8801,
+      "step": 170110
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.70809024022423e-05,
+      "loss": 1.1005,
+      "step": 170120
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.7080140752814295e-05,
+      "loss": 0.8761,
+      "step": 170130
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.70793791033863e-05,
+      "loss": 0.817,
+      "step": 170140
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707861745395829e-05,
+      "loss": 0.9456,
+      "step": 170150
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.7077855804530294e-05,
+      "loss": 0.9171,
+      "step": 170160
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707709415510229e-05,
+      "loss": 0.8972,
+      "step": 170170
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707633250567429e-05,
+      "loss": 0.8606,
+      "step": 170180
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.7075570856246286e-05,
+      "loss": 0.8255,
+      "step": 170190
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707480920681829e-05,
+      "loss": 1.0419,
+      "step": 170200
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707404755739028e-05,
+      "loss": 0.9471,
+      "step": 170210
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.7073285907962286e-05,
+      "loss": 0.9892,
+      "step": 170220
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707252425853428e-05,
+      "loss": 0.9089,
+      "step": 170230
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707176260910628e-05,
+      "loss": 0.9053,
+      "step": 170240
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.7071000959678285e-05,
+      "loss": 0.8548,
+      "step": 170250
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.707023931025028e-05,
+      "loss": 0.9055,
+      "step": 170260
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.706947766082228e-05,
+      "loss": 0.845,
+      "step": 170270
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706871601139428e-05,
+      "loss": 0.8391,
+      "step": 170280
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706795436196628e-05,
+      "loss": 0.9418,
+      "step": 170290
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706719271253827e-05,
+      "loss": 0.7574,
+      "step": 170300
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7066431063110276e-05,
+      "loss": 0.9585,
+      "step": 170310
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7065669413682274e-05,
+      "loss": 0.8901,
+      "step": 170320
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706490776425427e-05,
+      "loss": 0.9329,
+      "step": 170330
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706414611482627e-05,
+      "loss": 0.974,
+      "step": 170340
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706338446539827e-05,
+      "loss": 0.8365,
+      "step": 170350
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7062622815970264e-05,
+      "loss": 0.9572,
+      "step": 170360
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706186116654227e-05,
+      "loss": 0.8398,
+      "step": 170370
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7061099517114265e-05,
+      "loss": 1.0545,
+      "step": 170380
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.706033786768626e-05,
+      "loss": 0.9493,
+      "step": 170390
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.705957621825826e-05,
+      "loss": 1.048,
+      "step": 170400
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.705881456883026e-05,
+      "loss": 0.8188,
+      "step": 170410
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7058052919402255e-05,
+      "loss": 0.8411,
+      "step": 170420
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.705729126997426e-05,
+      "loss": 0.7894,
+      "step": 170430
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.705652962054626e-05,
+      "loss": 0.8172,
+      "step": 170440
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7055767971118254e-05,
+      "loss": 0.8692,
+      "step": 170450
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.705500632169026e-05,
+      "loss": 0.817,
+      "step": 170460
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.705424467226225e-05,
+      "loss": 0.8761,
+      "step": 170470
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7053483022834254e-05,
+      "loss": 0.8149,
+      "step": 170480
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.705272137340625e-05,
+      "loss": 0.9144,
+      "step": 170490
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.705195972397825e-05,
+      "loss": 0.9491,
+      "step": 170500
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.7051198074550246e-05,
+      "loss": 0.982,
+      "step": 170510
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.705043642512225e-05,
+      "loss": 1.0017,
+      "step": 170520
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704967477569424e-05,
+      "loss": 0.8501,
+      "step": 170530
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.7048913126266245e-05,
+      "loss": 0.9134,
+      "step": 170540
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704815147683824e-05,
+      "loss": 0.947,
+      "step": 170550
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704738982741024e-05,
+      "loss": 0.9481,
+      "step": 170560
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704662817798224e-05,
+      "loss": 0.8987,
+      "step": 170570
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704586652855424e-05,
+      "loss": 1.0025,
+      "step": 170580
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704510487912623e-05,
+      "loss": 0.8738,
+      "step": 170590
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704434322969824e-05,
+      "loss": 0.8719,
+      "step": 170600
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.7043581580270234e-05,
+      "loss": 0.8426,
+      "step": 170610
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704281993084223e-05,
+      "loss": 0.8718,
+      "step": 170620
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704205828141423e-05,
+      "loss": 0.9858,
+      "step": 170630
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.7041296631986233e-05,
+      "loss": 1.0044,
+      "step": 170640
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.704053498255823e-05,
+      "loss": 1.0186,
+      "step": 170650
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.703977333313023e-05,
+      "loss": 0.9017,
+      "step": 170660
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.703901168370223e-05,
+      "loss": 0.8735,
+      "step": 170670
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.703825003427422e-05,
+      "loss": 0.9769,
+      "step": 170680
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.703748838484623e-05,
+      "loss": 0.9594,
+      "step": 170690
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.7036726735418225e-05,
+      "loss": 0.7727,
+      "step": 170700
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.703596508599022e-05,
+      "loss": 0.9653,
+      "step": 170710
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.703520343656222e-05,
+      "loss": 0.9105,
+      "step": 170720
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7034441787134224e-05,
+      "loss": 0.9526,
+      "step": 170730
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7033680137706215e-05,
+      "loss": 0.916,
+      "step": 170740
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.703291848827822e-05,
+      "loss": 0.9726,
+      "step": 170750
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7032156838850217e-05,
+      "loss": 0.8896,
+      "step": 170760
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7031395189422214e-05,
+      "loss": 0.8372,
+      "step": 170770
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.703063353999421e-05,
+      "loss": 0.9681,
+      "step": 170780
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7029871890566216e-05,
+      "loss": 0.9709,
+      "step": 170790
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7029110241138206e-05,
+      "loss": 0.8119,
+      "step": 170800
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.702834859171021e-05,
+      "loss": 0.8221,
+      "step": 170810
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.702758694228221e-05,
+      "loss": 0.99,
+      "step": 170820
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7026825292854206e-05,
+      "loss": 1.0251,
+      "step": 170830
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.702606364342621e-05,
+      "loss": 0.883,
+      "step": 170840
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.70253019939982e-05,
+      "loss": 0.9521,
+      "step": 170850
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7024540344570205e-05,
+      "loss": 0.917,
+      "step": 170860
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.70237786951422e-05,
+      "loss": 0.9685,
+      "step": 170870
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.70230170457142e-05,
+      "loss": 0.9635,
+      "step": 170880
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.70222553962862e-05,
+      "loss": 0.9002,
+      "step": 170890
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.70214937468582e-05,
+      "loss": 0.8957,
+      "step": 170900
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.702073209743019e-05,
+      "loss": 0.9508,
+      "step": 170910
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.7019970448002196e-05,
+      "loss": 0.8449,
+      "step": 170920
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7019208798574194e-05,
+      "loss": 0.8378,
+      "step": 170930
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701844714914619e-05,
+      "loss": 0.8819,
+      "step": 170940
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701768549971819e-05,
+      "loss": 0.9109,
+      "step": 170950
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701692385029019e-05,
+      "loss": 0.907,
+      "step": 170960
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7016162200862184e-05,
+      "loss": 0.8731,
+      "step": 170970
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701540055143419e-05,
+      "loss": 0.905,
+      "step": 170980
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7014638902006185e-05,
+      "loss": 0.9906,
+      "step": 170990
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701387725257818e-05,
+      "loss": 0.858,
+      "step": 171000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701311560315018e-05,
+      "loss": 0.9366,
+      "step": 171010
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7012353953722185e-05,
+      "loss": 0.8469,
+      "step": 171020
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701159230429418e-05,
+      "loss": 0.8616,
+      "step": 171030
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.701083065486618e-05,
+      "loss": 0.845,
+      "step": 171040
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7010069005438184e-05,
+      "loss": 0.9581,
+      "step": 171050
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7009307356010174e-05,
+      "loss": 0.9279,
+      "step": 171060
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.700854570658218e-05,
+      "loss": 0.9172,
+      "step": 171070
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7007784057154176e-05,
+      "loss": 0.9036,
+      "step": 171080
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7007022407726174e-05,
+      "loss": 0.9049,
+      "step": 171090
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.700626075829817e-05,
+      "loss": 0.9463,
+      "step": 171100
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7005499108870175e-05,
+      "loss": 1.0245,
+      "step": 171110
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7004737459442166e-05,
+      "loss": 0.8852,
+      "step": 171120
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.700397581001417e-05,
+      "loss": 0.9513,
+      "step": 171130
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.700321416058617e-05,
+      "loss": 1.022,
+      "step": 171140
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.7002452511158165e-05,
+      "loss": 0.9409,
+      "step": 171150
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.700169086173016e-05,
+      "loss": 0.9536,
+      "step": 171160
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.700092921230217e-05,
+      "loss": 0.9912,
+      "step": 171170
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.700016756287416e-05,
+      "loss": 1.0342,
+      "step": 171180
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699940591344616e-05,
+      "loss": 0.9378,
+      "step": 171190
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699864426401816e-05,
+      "loss": 0.9672,
+      "step": 171200
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699788261459016e-05,
+      "loss": 0.9332,
+      "step": 171210
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.6997120965162154e-05,
+      "loss": 0.8909,
+      "step": 171220
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699635931573416e-05,
+      "loss": 0.9921,
+      "step": 171230
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.6995597666306156e-05,
+      "loss": 1.0295,
+      "step": 171240
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.6994836016878153e-05,
+      "loss": 0.9435,
+      "step": 171250
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699407436745016e-05,
+      "loss": 0.9683,
+      "step": 171260
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699331271802215e-05,
+      "loss": 0.8404,
+      "step": 171270
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699255106859415e-05,
+      "loss": 0.7878,
+      "step": 171280
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699178941916615e-05,
+      "loss": 0.9303,
+      "step": 171290
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.699102776973815e-05,
+      "loss": 1.0477,
+      "step": 171300
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.6990266120310145e-05,
+      "loss": 0.886,
+      "step": 171310
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.698950447088214e-05,
+      "loss": 0.8732,
+      "step": 171320
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.698874282145414e-05,
+      "loss": 0.9026,
+      "step": 171330
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.6987981172026144e-05,
+      "loss": 0.9382,
+      "step": 171340
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.6987219522598135e-05,
+      "loss": 1.0919,
+      "step": 171350
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.698645787317014e-05,
+      "loss": 0.8924,
+      "step": 171360
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6985696223742137e-05,
+      "loss": 0.9283,
+      "step": 171370
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6984934574314134e-05,
+      "loss": 0.8753,
+      "step": 171380
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.698417292488613e-05,
+      "loss": 0.947,
+      "step": 171390
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6983411275458136e-05,
+      "loss": 0.9828,
+      "step": 171400
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.698264962603013e-05,
+      "loss": 0.9655,
+      "step": 171410
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.698188797660213e-05,
+      "loss": 0.8623,
+      "step": 171420
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.698112632717413e-05,
+      "loss": 0.9134,
+      "step": 171430
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6980364677746126e-05,
+      "loss": 0.8634,
+      "step": 171440
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697960302831813e-05,
+      "loss": 0.9716,
+      "step": 171450
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697884137889013e-05,
+      "loss": 1.0011,
+      "step": 171460
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6978079729462125e-05,
+      "loss": 0.9075,
+      "step": 171470
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697731808003412e-05,
+      "loss": 0.8995,
+      "step": 171480
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6976556430606126e-05,
+      "loss": 0.9275,
+      "step": 171490
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697579478117812e-05,
+      "loss": 0.8553,
+      "step": 171500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697503313175012e-05,
+      "loss": 0.9726,
+      "step": 171510
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697427148232212e-05,
+      "loss": 0.8065,
+      "step": 171520
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6973509832894116e-05,
+      "loss": 0.9087,
+      "step": 171530
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.6972748183466114e-05,
+      "loss": 0.8216,
+      "step": 171540
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697198653403812e-05,
+      "loss": 0.8664,
+      "step": 171550
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697122488461011e-05,
+      "loss": 0.99,
+      "step": 171560
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.697046323518211e-05,
+      "loss": 0.8277,
+      "step": 171570
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.696970158575411e-05,
+      "loss": 1.1339,
+      "step": 171580
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.696893993632611e-05,
+      "loss": 0.9107,
+      "step": 171590
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.6968178286898105e-05,
+      "loss": 0.8495,
+      "step": 171600
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.696741663747011e-05,
+      "loss": 0.9315,
+      "step": 171610
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.696665498804211e-05,
+      "loss": 0.9014,
+      "step": 171620
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.6965893338614105e-05,
+      "loss": 0.9084,
+      "step": 171630
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.696513168918611e-05,
+      "loss": 0.9829,
+      "step": 171640
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.69643700397581e-05,
+      "loss": 0.8909,
+      "step": 171650
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.6963608390330104e-05,
+      "loss": 0.912,
+      "step": 171660
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.69628467409021e-05,
+      "loss": 1.0618,
+      "step": 171670
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.69620850914741e-05,
+      "loss": 0.9918,
+      "step": 171680
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.6961323442046096e-05,
+      "loss": 1.0419,
+      "step": 171690
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.69605617926181e-05,
+      "loss": 0.9408,
+      "step": 171700
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695980014319009e-05,
+      "loss": 0.893,
+      "step": 171710
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.6959038493762095e-05,
+      "loss": 0.9312,
+      "step": 171720
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695827684433409e-05,
+      "loss": 1.0283,
+      "step": 171730
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695751519490609e-05,
+      "loss": 0.8449,
+      "step": 171740
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695675354547809e-05,
+      "loss": 1.1677,
+      "step": 171750
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695599189605009e-05,
+      "loss": 1.0426,
+      "step": 171760
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695523024662208e-05,
+      "loss": 0.7619,
+      "step": 171770
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695446859719409e-05,
+      "loss": 0.8578,
+      "step": 171780
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.6953706947766084e-05,
+      "loss": 0.9432,
+      "step": 171790
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.695294529833808e-05,
+      "loss": 0.8925,
+      "step": 171800
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.695218364891008e-05,
+      "loss": 0.8792,
+      "step": 171810
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.695142199948208e-05,
+      "loss": 0.9248,
+      "step": 171820
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.695066035005408e-05,
+      "loss": 0.8884,
+      "step": 171830
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694989870062608e-05,
+      "loss": 1.003,
+      "step": 171840
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.6949137051198076e-05,
+      "loss": 1.0498,
+      "step": 171850
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.6948375401770073e-05,
+      "loss": 0.948,
+      "step": 171860
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694761375234208e-05,
+      "loss": 0.787,
+      "step": 171870
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694685210291407e-05,
+      "loss": 0.9318,
+      "step": 171880
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694609045348607e-05,
+      "loss": 0.8571,
+      "step": 171890
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694532880405807e-05,
+      "loss": 0.8765,
+      "step": 171900
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694456715463007e-05,
+      "loss": 0.9195,
+      "step": 171910
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.6943805505202065e-05,
+      "loss": 0.9564,
+      "step": 171920
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694304385577407e-05,
+      "loss": 0.9584,
+      "step": 171930
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694228220634606e-05,
+      "loss": 1.0173,
+      "step": 171940
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.6941520556918064e-05,
+      "loss": 0.9169,
+      "step": 171950
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.694075890749006e-05,
+      "loss": 0.9698,
+      "step": 171960
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.693999725806206e-05,
+      "loss": 0.9047,
+      "step": 171970
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.6939235608634057e-05,
+      "loss": 0.7819,
+      "step": 171980
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.693847395920606e-05,
+      "loss": 0.8752,
+      "step": 171990
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.693771230977806e-05,
+      "loss": 0.9421,
+      "step": 172000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.6936950660350056e-05,
+      "loss": 0.9198,
+      "step": 172010
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.693618901092205e-05,
+      "loss": 0.9217,
+      "step": 172020
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.693542736149405e-05,
+      "loss": 0.8762,
+      "step": 172030
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6934665712066055e-05,
+      "loss": 0.947,
+      "step": 172040
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.693390406263805e-05,
+      "loss": 0.8972,
+      "step": 172050
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.693314241321005e-05,
+      "loss": 0.8455,
+      "step": 172060
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.693238076378205e-05,
+      "loss": 0.8804,
+      "step": 172070
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.693161911435405e-05,
+      "loss": 0.9025,
+      "step": 172080
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.693085746492604e-05,
+      "loss": 0.8928,
+      "step": 172090
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6930095815498047e-05,
+      "loss": 0.9711,
+      "step": 172100
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6929334166070044e-05,
+      "loss": 0.8743,
+      "step": 172110
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692857251664204e-05,
+      "loss": 0.9711,
+      "step": 172120
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692781086721404e-05,
+      "loss": 0.864,
+      "step": 172130
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692704921778604e-05,
+      "loss": 0.8953,
+      "step": 172140
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6926287568358034e-05,
+      "loss": 0.8661,
+      "step": 172150
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692552591893004e-05,
+      "loss": 0.9132,
+      "step": 172160
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6924764269502036e-05,
+      "loss": 0.9612,
+      "step": 172170
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692400262007403e-05,
+      "loss": 0.9811,
+      "step": 172180
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692324097064603e-05,
+      "loss": 0.8519,
+      "step": 172190
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6922479321218035e-05,
+      "loss": 0.8784,
+      "step": 172200
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692171767179003e-05,
+      "loss": 1.0034,
+      "step": 172210
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692095602236203e-05,
+      "loss": 0.8767,
+      "step": 172220
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.692019437293403e-05,
+      "loss": 0.8826,
+      "step": 172230
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.6919432723506025e-05,
+      "loss": 0.8831,
+      "step": 172240
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691867107407803e-05,
+      "loss": 0.9648,
+      "step": 172250
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691790942465002e-05,
+      "loss": 0.8146,
+      "step": 172260
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6917147775222024e-05,
+      "loss": 0.907,
+      "step": 172270
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691638612579402e-05,
+      "loss": 0.7815,
+      "step": 172280
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691562447636602e-05,
+      "loss": 0.861,
+      "step": 172290
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6914862826938016e-05,
+      "loss": 1.04,
+      "step": 172300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691410117751002e-05,
+      "loss": 0.9158,
+      "step": 172310
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691333952808201e-05,
+      "loss": 0.9507,
+      "step": 172320
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6912577878654015e-05,
+      "loss": 0.9081,
+      "step": 172330
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691181622922601e-05,
+      "loss": 0.9768,
+      "step": 172340
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691105457979801e-05,
+      "loss": 0.8914,
+      "step": 172350
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.691029293037001e-05,
+      "loss": 0.8717,
+      "step": 172360
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.690953128094201e-05,
+      "loss": 0.9117,
+      "step": 172370
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.690876963151401e-05,
+      "loss": 0.8783,
+      "step": 172380
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.690800798208601e-05,
+      "loss": 0.9073,
+      "step": 172390
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6907246332658004e-05,
+      "loss": 0.9319,
+      "step": 172400
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.690648468323e-05,
+      "loss": 0.8384,
+      "step": 172410
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6905723033802006e-05,
+      "loss": 0.9841,
+      "step": 172420
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6904961384374004e-05,
+      "loss": 0.8886,
+      "step": 172430
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6904199734946e-05,
+      "loss": 0.9192,
+      "step": 172440
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.6903438085518e-05,
+      "loss": 0.7156,
+      "step": 172450
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.690267643609e-05,
+      "loss": 0.9018,
+      "step": 172460
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6901914786661993e-05,
+      "loss": 1.0043,
+      "step": 172470
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6901153137234e-05,
+      "loss": 0.9319,
+      "step": 172480
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6900391487805995e-05,
+      "loss": 0.9421,
+      "step": 172490
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689962983837799e-05,
+      "loss": 0.9504,
+      "step": 172500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689886818894999e-05,
+      "loss": 0.9545,
+      "step": 172510
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6898106539521994e-05,
+      "loss": 0.9032,
+      "step": 172520
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6897344890093985e-05,
+      "loss": 0.8498,
+      "step": 172530
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689658324066599e-05,
+      "loss": 0.8622,
+      "step": 172540
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689582159123799e-05,
+      "loss": 0.9364,
+      "step": 172550
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6895059941809984e-05,
+      "loss": 0.84,
+      "step": 172560
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689429829238198e-05,
+      "loss": 0.9103,
+      "step": 172570
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6893536642953986e-05,
+      "loss": 0.8861,
+      "step": 172580
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689277499352598e-05,
+      "loss": 0.9472,
+      "step": 172590
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689201334409798e-05,
+      "loss": 0.8637,
+      "step": 172600
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.689125169466998e-05,
+      "loss": 0.9614,
+      "step": 172610
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6890490045241976e-05,
+      "loss": 0.9089,
+      "step": 172620
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.688972839581398e-05,
+      "loss": 0.9813,
+      "step": 172630
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.688896674638598e-05,
+      "loss": 0.8791,
+      "step": 172640
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.6888205096957975e-05,
+      "loss": 0.8203,
+      "step": 172650
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.688744344752997e-05,
+      "loss": 1.0031,
+      "step": 172660
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.688668179810198e-05,
+      "loss": 0.849,
+      "step": 172670
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.688592014867397e-05,
+      "loss": 0.8602,
+      "step": 172680
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.688515849924597e-05,
+      "loss": 1.0546,
+      "step": 172690
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.688439684981796e-05,
+      "loss": 0.8529,
+      "step": 172700
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6883635200389967e-05,
+      "loss": 0.9245,
+      "step": 172710
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6882873550961964e-05,
+      "loss": 0.8954,
+      "step": 172720
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.688211190153396e-05,
+      "loss": 0.9019,
+      "step": 172730
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.688135025210596e-05,
+      "loss": 0.9085,
+      "step": 172740
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.688058860267796e-05,
+      "loss": 0.9434,
+      "step": 172750
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687982695324996e-05,
+      "loss": 0.9081,
+      "step": 172760
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687906530382196e-05,
+      "loss": 0.9823,
+      "step": 172770
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6878303654393956e-05,
+      "loss": 0.9571,
+      "step": 172780
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687754200496595e-05,
+      "loss": 0.8755,
+      "step": 172790
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687678035553796e-05,
+      "loss": 0.8472,
+      "step": 172800
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6876018706109955e-05,
+      "loss": 1.0111,
+      "step": 172810
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687525705668195e-05,
+      "loss": 0.9803,
+      "step": 172820
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687449540725395e-05,
+      "loss": 0.8095,
+      "step": 172830
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6873733757825954e-05,
+      "loss": 0.9536,
+      "step": 172840
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6872972108397945e-05,
+      "loss": 0.9744,
+      "step": 172850
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.687221045896995e-05,
+      "loss": 0.8808,
+      "step": 172860
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6871448809541946e-05,
+      "loss": 0.9414,
+      "step": 172870
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.6870687160113944e-05,
+      "loss": 0.898,
+      "step": 172880
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.686992551068594e-05,
+      "loss": 0.8821,
+      "step": 172890
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6869163861257945e-05,
+      "loss": 0.8462,
+      "step": 172900
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6868402211829936e-05,
+      "loss": 0.8108,
+      "step": 172910
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686764056240194e-05,
+      "loss": 0.9292,
+      "step": 172920
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686687891297394e-05,
+      "loss": 0.9263,
+      "step": 172930
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6866117263545935e-05,
+      "loss": 0.9994,
+      "step": 172940
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686535561411793e-05,
+      "loss": 0.7812,
+      "step": 172950
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686459396468994e-05,
+      "loss": 0.8906,
+      "step": 172960
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6863832315261935e-05,
+      "loss": 0.8301,
+      "step": 172970
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686307066583393e-05,
+      "loss": 0.8169,
+      "step": 172980
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686230901640593e-05,
+      "loss": 0.7543,
+      "step": 172990
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686154736697793e-05,
+      "loss": 1.0165,
+      "step": 173000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686078571754993e-05,
+      "loss": 1.1058,
+      "step": 173010
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.686002406812193e-05,
+      "loss": 0.898,
+      "step": 173020
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6859262418693926e-05,
+      "loss": 0.8607,
+      "step": 173030
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6858500769265924e-05,
+      "loss": 0.8675,
+      "step": 173040
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.685773911983793e-05,
+      "loss": 0.967,
+      "step": 173050
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.685697747040992e-05,
+      "loss": 0.9772,
+      "step": 173060
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.685621582098192e-05,
+      "loss": 0.8501,
+      "step": 173070
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.685545417155392e-05,
+      "loss": 0.9226,
+      "step": 173080
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.685469252212592e-05,
+      "loss": 0.8396,
+      "step": 173090
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.6853930872697915e-05,
+      "loss": 0.9927,
+      "step": 173100
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.685316922326992e-05,
+      "loss": 0.8084,
+      "step": 173110
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.685240757384191e-05,
+      "loss": 0.914,
+      "step": 173120
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6851645924413914e-05,
+      "loss": 0.9116,
+      "step": 173130
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.685088427498591e-05,
+      "loss": 0.8654,
+      "step": 173140
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.685012262555791e-05,
+      "loss": 0.9248,
+      "step": 173150
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.684936097612991e-05,
+      "loss": 0.8955,
+      "step": 173160
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6848599326701904e-05,
+      "loss": 0.9935,
+      "step": 173170
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.684783767727391e-05,
+      "loss": 0.8924,
+      "step": 173180
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6847076027845906e-05,
+      "loss": 1.0453,
+      "step": 173190
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.68463143784179e-05,
+      "loss": 0.955,
+      "step": 173200
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.68455527289899e-05,
+      "loss": 0.8345,
+      "step": 173210
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6844791079561905e-05,
+      "loss": 0.8002,
+      "step": 173220
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6844029430133896e-05,
+      "loss": 1.0521,
+      "step": 173230
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.68432677807059e-05,
+      "loss": 0.9059,
+      "step": 173240
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.68425061312779e-05,
+      "loss": 0.9683,
+      "step": 173250
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6841744481849895e-05,
+      "loss": 0.9348,
+      "step": 173260
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.684098283242189e-05,
+      "loss": 0.964,
+      "step": 173270
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.68402211829939e-05,
+      "loss": 0.8696,
+      "step": 173280
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.683945953356589e-05,
+      "loss": 0.8713,
+      "step": 173290
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.683869788413789e-05,
+      "loss": 0.8827,
+      "step": 173300
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.683793623470989e-05,
+      "loss": 0.9327,
+      "step": 173310
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6837174585281887e-05,
+      "loss": 0.883,
+      "step": 173320
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.6836412935853884e-05,
+      "loss": 0.9198,
+      "step": 173330
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683565128642589e-05,
+      "loss": 0.8841,
+      "step": 173340
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.6834889636997886e-05,
+      "loss": 0.9215,
+      "step": 173350
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683412798756988e-05,
+      "loss": 0.9458,
+      "step": 173360
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683336633814188e-05,
+      "loss": 0.9254,
+      "step": 173370
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683260468871388e-05,
+      "loss": 0.884,
+      "step": 173380
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683184303928588e-05,
+      "loss": 1.056,
+      "step": 173390
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683108138985788e-05,
+      "loss": 0.9711,
+      "step": 173400
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.683031974042988e-05,
+      "loss": 0.9762,
+      "step": 173410
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.6829558091001875e-05,
+      "loss": 0.9147,
+      "step": 173420
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682879644157388e-05,
+      "loss": 1.0318,
+      "step": 173430
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682803479214587e-05,
+      "loss": 0.9136,
+      "step": 173440
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.6827273142717874e-05,
+      "loss": 1.0904,
+      "step": 173450
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682651149328987e-05,
+      "loss": 0.9936,
+      "step": 173460
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682574984386187e-05,
+      "loss": 0.9431,
+      "step": 173470
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.6824988194433866e-05,
+      "loss": 0.9034,
+      "step": 173480
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682422654500587e-05,
+      "loss": 0.9325,
+      "step": 173490
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682346489557786e-05,
+      "loss": 0.9542,
+      "step": 173500
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.6822703246149865e-05,
+      "loss": 0.8711,
+      "step": 173510
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682194159672186e-05,
+      "loss": 0.8927,
+      "step": 173520
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682117994729386e-05,
+      "loss": 0.7966,
+      "step": 173530
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.682041829786586e-05,
+      "loss": 0.9244,
+      "step": 173540
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.681965664843786e-05,
+      "loss": 0.946,
+      "step": 173550
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681889499900986e-05,
+      "loss": 0.8147,
+      "step": 173560
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681813334958186e-05,
+      "loss": 0.9152,
+      "step": 173570
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6817371700153855e-05,
+      "loss": 0.8867,
+      "step": 173580
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681661005072585e-05,
+      "loss": 0.9567,
+      "step": 173590
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6815848401297856e-05,
+      "loss": 0.8942,
+      "step": 173600
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681508675186985e-05,
+      "loss": 0.9257,
+      "step": 173610
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681432510244185e-05,
+      "loss": 0.9067,
+      "step": 173620
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681356345301385e-05,
+      "loss": 0.8966,
+      "step": 173630
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6812801803585846e-05,
+      "loss": 0.9024,
+      "step": 173640
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6812040154157844e-05,
+      "loss": 0.8616,
+      "step": 173650
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681127850472985e-05,
+      "loss": 0.9004,
+      "step": 173660
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.681051685530184e-05,
+      "loss": 0.8701,
+      "step": 173670
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680975520587384e-05,
+      "loss": 1.0111,
+      "step": 173680
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680899355644584e-05,
+      "loss": 0.919,
+      "step": 173690
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680823190701784e-05,
+      "loss": 0.868,
+      "step": 173700
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6807470257589835e-05,
+      "loss": 0.9697,
+      "step": 173710
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680670860816184e-05,
+      "loss": 0.8722,
+      "step": 173720
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680594695873383e-05,
+      "loss": 0.7421,
+      "step": 173730
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6805185309305834e-05,
+      "loss": 0.853,
+      "step": 173740
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680442365987783e-05,
+      "loss": 0.8912,
+      "step": 173750
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.680366201044983e-05,
+      "loss": 1.1429,
+      "step": 173760
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.6802900361021834e-05,
+      "loss": 0.9435,
+      "step": 173770
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.680213871159383e-05,
+      "loss": 0.9261,
+      "step": 173780
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.680137706216583e-05,
+      "loss": 0.9491,
+      "step": 173790
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.6800615412737826e-05,
+      "loss": 0.9186,
+      "step": 173800
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679985376330983e-05,
+      "loss": 0.9511,
+      "step": 173810
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679909211388182e-05,
+      "loss": 0.9969,
+      "step": 173820
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.6798330464453825e-05,
+      "loss": 0.9978,
+      "step": 173830
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679756881502582e-05,
+      "loss": 0.9216,
+      "step": 173840
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679680716559782e-05,
+      "loss": 0.9505,
+      "step": 173850
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679604551616982e-05,
+      "loss": 0.9032,
+      "step": 173860
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679528386674182e-05,
+      "loss": 0.8256,
+      "step": 173870
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679452221731381e-05,
+      "loss": 0.9073,
+      "step": 173880
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679376056788582e-05,
+      "loss": 0.9235,
+      "step": 173890
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.6792998918457814e-05,
+      "loss": 0.9634,
+      "step": 173900
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679223726902981e-05,
+      "loss": 1.0189,
+      "step": 173910
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679147561960181e-05,
+      "loss": 0.8867,
+      "step": 173920
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.679071397017381e-05,
+      "loss": 0.8765,
+      "step": 173930
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.678995232074581e-05,
+      "loss": 0.8941,
+      "step": 173940
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.678919067131781e-05,
+      "loss": 0.8378,
+      "step": 173950
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.6788429021889806e-05,
+      "loss": 1.0035,
+      "step": 173960
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.67876673724618e-05,
+      "loss": 0.8592,
+      "step": 173970
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.678690572303381e-05,
+      "loss": 1.0075,
+      "step": 173980
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.6786144073605805e-05,
+      "loss": 0.8301,
+      "step": 173990
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.67853824241778e-05,
+      "loss": 0.8407,
+      "step": 174000
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.67846207747498e-05,
+      "loss": 0.9291,
+      "step": 174010
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6783859125321804e-05,
+      "loss": 0.9056,
+      "step": 174020
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6783097475893795e-05,
+      "loss": 1.0621,
+      "step": 174030
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.67823358264658e-05,
+      "loss": 0.8933,
+      "step": 174040
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6781574177037796e-05,
+      "loss": 0.8865,
+      "step": 174050
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6780812527609794e-05,
+      "loss": 0.9626,
+      "step": 174060
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.678005087818179e-05,
+      "loss": 0.8819,
+      "step": 174070
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677928922875379e-05,
+      "loss": 0.9157,
+      "step": 174080
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6778527579325786e-05,
+      "loss": 0.9767,
+      "step": 174090
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677776592989779e-05,
+      "loss": 0.847,
+      "step": 174100
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677700428046978e-05,
+      "loss": 0.8759,
+      "step": 174110
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6776242631041786e-05,
+      "loss": 0.9094,
+      "step": 174120
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677548098161378e-05,
+      "loss": 0.914,
+      "step": 174130
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677471933218578e-05,
+      "loss": 0.959,
+      "step": 174140
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6773957682757785e-05,
+      "loss": 0.9185,
+      "step": 174150
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677319603332978e-05,
+      "loss": 0.7989,
+      "step": 174160
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677243438390178e-05,
+      "loss": 0.8138,
+      "step": 174170
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677167273447378e-05,
+      "loss": 0.8762,
+      "step": 174180
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677091108504578e-05,
+      "loss": 0.8557,
+      "step": 174190
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.677014943561777e-05,
+      "loss": 0.9107,
+      "step": 174200
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.6769387786189776e-05,
+      "loss": 0.9835,
+      "step": 174210
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6768626136761774e-05,
+      "loss": 0.9285,
+      "step": 174220
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676786448733377e-05,
+      "loss": 0.9735,
+      "step": 174230
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676710283790577e-05,
+      "loss": 0.9655,
+      "step": 174240
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676634118847777e-05,
+      "loss": 0.8574,
+      "step": 174250
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6765579539049764e-05,
+      "loss": 0.8789,
+      "step": 174260
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676481788962177e-05,
+      "loss": 0.8426,
+      "step": 174270
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6764056240193765e-05,
+      "loss": 0.9005,
+      "step": 174280
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676329459076576e-05,
+      "loss": 0.9462,
+      "step": 174290
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676253294133776e-05,
+      "loss": 0.8504,
+      "step": 174300
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6761771291909764e-05,
+      "loss": 0.969,
+      "step": 174310
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6761009642481755e-05,
+      "loss": 0.8884,
+      "step": 174320
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.676024799305376e-05,
+      "loss": 0.9547,
+      "step": 174330
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.675948634362576e-05,
+      "loss": 1.0078,
+      "step": 174340
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6758724694197754e-05,
+      "loss": 0.9932,
+      "step": 174350
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.675796304476976e-05,
+      "loss": 0.877,
+      "step": 174360
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6757201395341756e-05,
+      "loss": 0.961,
+      "step": 174370
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6756439745913754e-05,
+      "loss": 0.944,
+      "step": 174380
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.675567809648575e-05,
+      "loss": 0.974,
+      "step": 174390
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6754916447057755e-05,
+      "loss": 0.9882,
+      "step": 174400
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6754154797629746e-05,
+      "loss": 0.9853,
+      "step": 174410
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.675339314820175e-05,
+      "loss": 0.9145,
+      "step": 174420
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.675263149877375e-05,
+      "loss": 0.9712,
+      "step": 174430
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.6751869849345745e-05,
+      "loss": 0.8521,
+      "step": 174440
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.675110819991774e-05,
+      "loss": 0.9027,
+      "step": 174450
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.675034655048975e-05,
+      "loss": 0.8644,
+      "step": 174460
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674958490106174e-05,
+      "loss": 0.8907,
+      "step": 174470
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674882325163374e-05,
+      "loss": 1.0636,
+      "step": 174480
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674806160220574e-05,
+      "loss": 0.9384,
+      "step": 174490
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674729995277774e-05,
+      "loss": 0.9468,
+      "step": 174500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.6746538303349734e-05,
+      "loss": 0.9209,
+      "step": 174510
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674577665392174e-05,
+      "loss": 1.0343,
+      "step": 174520
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674501500449373e-05,
+      "loss": 0.9312,
+      "step": 174530
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674425335506573e-05,
+      "loss": 0.9152,
+      "step": 174540
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674349170563773e-05,
+      "loss": 0.864,
+      "step": 174550
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674273005620973e-05,
+      "loss": 1.0371,
+      "step": 174560
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674196840678173e-05,
+      "loss": 0.8982,
+      "step": 174570
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674120675735372e-05,
+      "loss": 0.9524,
+      "step": 174580
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.674044510792573e-05,
+      "loss": 0.9004,
+      "step": 174590
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.6739683458497725e-05,
+      "loss": 0.9639,
+      "step": 174600
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.673892180906972e-05,
+      "loss": 0.8341,
+      "step": 174610
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.673816015964172e-05,
+      "loss": 0.9101,
+      "step": 174620
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.6737398510213724e-05,
+      "loss": 0.8749,
+      "step": 174630
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.6736636860785715e-05,
+      "loss": 0.9071,
+      "step": 174640
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.673587521135772e-05,
+      "loss": 0.9114,
+      "step": 174650
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6735113561929716e-05,
+      "loss": 0.8981,
+      "step": 174660
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6734351912501714e-05,
+      "loss": 0.9492,
+      "step": 174670
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.673359026307371e-05,
+      "loss": 0.9117,
+      "step": 174680
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6732828613645716e-05,
+      "loss": 1.0683,
+      "step": 174690
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6732066964217706e-05,
+      "loss": 0.8387,
+      "step": 174700
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.673130531478971e-05,
+      "loss": 0.7978,
+      "step": 174710
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.673054366536171e-05,
+      "loss": 0.9744,
+      "step": 174720
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6729782015933706e-05,
+      "loss": 0.9312,
+      "step": 174730
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.672902036650571e-05,
+      "loss": 0.9764,
+      "step": 174740
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.672825871707771e-05,
+      "loss": 0.9881,
+      "step": 174750
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6727497067649705e-05,
+      "loss": 0.9962,
+      "step": 174760
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.67267354182217e-05,
+      "loss": 0.8683,
+      "step": 174770
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6725973768793706e-05,
+      "loss": 0.9587,
+      "step": 174780
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.67252121193657e-05,
+      "loss": 0.9015,
+      "step": 174790
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.67244504699377e-05,
+      "loss": 0.7877,
+      "step": 174800
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.67236888205097e-05,
+      "loss": 0.9617,
+      "step": 174810
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6722927171081696e-05,
+      "loss": 0.9025,
+      "step": 174820
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.6722165521653694e-05,
+      "loss": 1.0585,
+      "step": 174830
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.67214038722257e-05,
+      "loss": 0.9639,
+      "step": 174840
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.672064222279769e-05,
+      "loss": 0.9213,
+      "step": 174850
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.671988057336969e-05,
+      "loss": 0.8309,
+      "step": 174860
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671911892394169e-05,
+      "loss": 0.8805,
+      "step": 174870
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671835727451369e-05,
+      "loss": 1.0049,
+      "step": 174880
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6717595625085685e-05,
+      "loss": 0.9377,
+      "step": 174890
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671683397565769e-05,
+      "loss": 0.8371,
+      "step": 174900
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671607232622968e-05,
+      "loss": 0.833,
+      "step": 174910
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6715310676801684e-05,
+      "loss": 0.8874,
+      "step": 174920
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671454902737368e-05,
+      "loss": 0.9596,
+      "step": 174930
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671378737794568e-05,
+      "loss": 0.8858,
+      "step": 174940
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6713025728517684e-05,
+      "loss": 0.8513,
+      "step": 174950
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671226407908968e-05,
+      "loss": 0.9313,
+      "step": 174960
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.671150242966168e-05,
+      "loss": 0.8529,
+      "step": 174970
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6710740780233676e-05,
+      "loss": 0.9379,
+      "step": 174980
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670997913080568e-05,
+      "loss": 0.9571,
+      "step": 174990
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670921748137767e-05,
+      "loss": 1.0021,
+      "step": 175000
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6708455831949675e-05,
+      "loss": 0.8746,
+      "step": 175010
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6707694182521666e-05,
+      "loss": 0.924,
+      "step": 175020
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670693253309367e-05,
+      "loss": 0.9129,
+      "step": 175030
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670617088366567e-05,
+      "loss": 0.865,
+      "step": 175040
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.6705409234237665e-05,
+      "loss": 1.0883,
+      "step": 175050
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670464758480966e-05,
+      "loss": 0.8966,
+      "step": 175060
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670388593538167e-05,
+      "loss": 1.0283,
+      "step": 175070
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.670312428595366e-05,
+      "loss": 0.8908,
+      "step": 175080
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.670236263652566e-05,
+      "loss": 0.9802,
+      "step": 175090
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.670160098709766e-05,
+      "loss": 0.8892,
+      "step": 175100
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.670083933766966e-05,
+      "loss": 0.9482,
+      "step": 175110
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.6700077688241654e-05,
+      "loss": 0.8887,
+      "step": 175120
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669931603881366e-05,
+      "loss": 0.925,
+      "step": 175130
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.6698554389385656e-05,
+      "loss": 0.9385,
+      "step": 175140
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669779273995765e-05,
+      "loss": 1.0,
+      "step": 175150
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669703109052966e-05,
+      "loss": 0.8935,
+      "step": 175160
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669626944110165e-05,
+      "loss": 0.9373,
+      "step": 175170
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669550779167365e-05,
+      "loss": 0.8309,
+      "step": 175180
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669474614224565e-05,
+      "loss": 0.9303,
+      "step": 175190
+    },
+    {
+      "epoch": 8.0,
+      "eval_cer": 0.621710237010309,
+      "eval_em": 0.007473982970671713,
+      "eval_f1": 0.007473982970671713,
+      "eval_loss": 0.8942475914955139,
+      "eval_runtime": 1010.326,
+      "eval_samples_per_second": 10.462,
+      "eval_steps_per_second": 1.308,
+      "eval_wer": 0.9925260170293283,
+      "step": 175198
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669398449281765e-05,
+      "loss": 0.8545,
+      "step": 175200
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.6693222843389645e-05,
+      "loss": 0.8341,
+      "step": 175210
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669246119396165e-05,
+      "loss": 0.9841,
+      "step": 175220
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669169954453364e-05,
+      "loss": 0.9641,
+      "step": 175230
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.6690937895105644e-05,
+      "loss": 0.8981,
+      "step": 175240
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.669017624567764e-05,
+      "loss": 0.9457,
+      "step": 175250
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.668941459624964e-05,
+      "loss": 0.8567,
+      "step": 175260
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.6688652946821636e-05,
+      "loss": 0.8682,
+      "step": 175270
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.668789129739364e-05,
+      "loss": 0.8621,
+      "step": 175280
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.668712964796563e-05,
+      "loss": 0.7917,
+      "step": 175290
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.6686367998537636e-05,
+      "loss": 0.8585,
+      "step": 175300
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668560634910963e-05,
+      "loss": 0.8144,
+      "step": 175310
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668484469968163e-05,
+      "loss": 0.8223,
+      "step": 175320
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.6684083050253635e-05,
+      "loss": 0.9158,
+      "step": 175330
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668332140082563e-05,
+      "loss": 0.9451,
+      "step": 175340
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668255975139763e-05,
+      "loss": 0.8461,
+      "step": 175350
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668179810196963e-05,
+      "loss": 0.828,
+      "step": 175360
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668103645254163e-05,
+      "loss": 0.8189,
+      "step": 175370
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.668027480311362e-05,
+      "loss": 0.9109,
+      "step": 175380
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.6679513153685626e-05,
+      "loss": 0.9985,
+      "step": 175390
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.6678751504257624e-05,
+      "loss": 0.914,
+      "step": 175400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667798985482962e-05,
+      "loss": 0.8804,
+      "step": 175410
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667722820540162e-05,
+      "loss": 0.8581,
+      "step": 175420
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667646655597362e-05,
+      "loss": 0.8514,
+      "step": 175430
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.6675704906545614e-05,
+      "loss": 0.9323,
+      "step": 175440
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667494325711762e-05,
+      "loss": 0.7966,
+      "step": 175450
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667418160768961e-05,
+      "loss": 0.9207,
+      "step": 175460
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667341995826161e-05,
+      "loss": 0.924,
+      "step": 175470
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667265830883361e-05,
+      "loss": 0.9429,
+      "step": 175480
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667189665940561e-05,
+      "loss": 0.9173,
+      "step": 175490
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.6671135009977605e-05,
+      "loss": 0.7981,
+      "step": 175500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.667037336054961e-05,
+      "loss": 0.8917,
+      "step": 175510
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.666961171112161e-05,
+      "loss": 0.8681,
+      "step": 175520
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6668850061693605e-05,
+      "loss": 0.8844,
+      "step": 175530
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.666808841226561e-05,
+      "loss": 0.8778,
+      "step": 175540
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.66673267628376e-05,
+      "loss": 0.7451,
+      "step": 175550
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6666565113409604e-05,
+      "loss": 0.7695,
+      "step": 175560
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.66658034639816e-05,
+      "loss": 0.8008,
+      "step": 175570
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.66650418145536e-05,
+      "loss": 0.8788,
+      "step": 175580
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6664280165125596e-05,
+      "loss": 0.8638,
+      "step": 175590
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.66635185156976e-05,
+      "loss": 0.9525,
+      "step": 175600
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.666275686626959e-05,
+      "loss": 0.9721,
+      "step": 175610
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6661995216841595e-05,
+      "loss": 0.8846,
+      "step": 175620
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.666123356741359e-05,
+      "loss": 0.9205,
+      "step": 175630
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.666047191798559e-05,
+      "loss": 0.8877,
+      "step": 175640
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665971026855759e-05,
+      "loss": 0.8716,
+      "step": 175650
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665894861912959e-05,
+      "loss": 0.8967,
+      "step": 175660
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665818696970158e-05,
+      "loss": 0.8664,
+      "step": 175670
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665742532027359e-05,
+      "loss": 0.794,
+      "step": 175680
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6656663670845584e-05,
+      "loss": 0.9194,
+      "step": 175690
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665590202141758e-05,
+      "loss": 0.8254,
+      "step": 175700
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665514037198958e-05,
+      "loss": 0.8607,
+      "step": 175710
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6654378722561583e-05,
+      "loss": 0.81,
+      "step": 175720
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665361707313358e-05,
+      "loss": 0.9329,
+      "step": 175730
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.665285542370558e-05,
+      "loss": 0.9159,
+      "step": 175740
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.665209377427758e-05,
+      "loss": 0.877,
+      "step": 175750
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.665133212484957e-05,
+      "loss": 0.901,
+      "step": 175760
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.665057047542158e-05,
+      "loss": 0.9721,
+      "step": 175770
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6649808825993575e-05,
+      "loss": 0.8955,
+      "step": 175780
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664904717656557e-05,
+      "loss": 0.8709,
+      "step": 175790
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664828552713757e-05,
+      "loss": 0.9171,
+      "step": 175800
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6647523877709574e-05,
+      "loss": 0.9221,
+      "step": 175810
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6646762228281565e-05,
+      "loss": 0.8684,
+      "step": 175820
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664600057885357e-05,
+      "loss": 0.8247,
+      "step": 175830
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664523892942557e-05,
+      "loss": 0.836,
+      "step": 175840
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6644477279997564e-05,
+      "loss": 0.8959,
+      "step": 175850
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664371563056956e-05,
+      "loss": 0.9445,
+      "step": 175860
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6642953981141566e-05,
+      "loss": 0.9869,
+      "step": 175870
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6642192331713556e-05,
+      "loss": 0.9139,
+      "step": 175880
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664143068228556e-05,
+      "loss": 0.8291,
+      "step": 175890
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.664066903285756e-05,
+      "loss": 0.8861,
+      "step": 175900
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6639907383429556e-05,
+      "loss": 0.9895,
+      "step": 175910
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.663914573400155e-05,
+      "loss": 0.8789,
+      "step": 175920
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.663838408457355e-05,
+      "loss": 0.9535,
+      "step": 175930
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.6637622435145555e-05,
+      "loss": 0.8676,
+      "step": 175940
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.663686078571755e-05,
+      "loss": 0.8665,
+      "step": 175950
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.663609913628955e-05,
+      "loss": 0.8996,
+      "step": 175960
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.663533748686155e-05,
+      "loss": 0.9119,
+      "step": 175970
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.663457583743355e-05,
+      "loss": 0.9703,
+      "step": 175980
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.663381418800554e-05,
+      "loss": 0.8692,
+      "step": 175990
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6633052538577546e-05,
+      "loss": 1.0017,
+      "step": 176000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6632290889149544e-05,
+      "loss": 0.9014,
+      "step": 176010
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.663152923972154e-05,
+      "loss": 0.8363,
+      "step": 176020
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.663076759029354e-05,
+      "loss": 0.9284,
+      "step": 176030
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.663000594086554e-05,
+      "loss": 0.8813,
+      "step": 176040
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6629244291437534e-05,
+      "loss": 0.9085,
+      "step": 176050
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.662848264200954e-05,
+      "loss": 0.8998,
+      "step": 176060
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6627720992581535e-05,
+      "loss": 0.8609,
+      "step": 176070
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.662695934315353e-05,
+      "loss": 0.8381,
+      "step": 176080
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.662619769372553e-05,
+      "loss": 0.8246,
+      "step": 176090
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6625436044297535e-05,
+      "loss": 0.9278,
+      "step": 176100
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.662467439486953e-05,
+      "loss": 1.0157,
+      "step": 176110
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.662391274544153e-05,
+      "loss": 0.8966,
+      "step": 176120
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6623151096013534e-05,
+      "loss": 0.9902,
+      "step": 176130
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6622389446585525e-05,
+      "loss": 0.9598,
+      "step": 176140
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.662162779715753e-05,
+      "loss": 0.9172,
+      "step": 176150
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6620866147729526e-05,
+      "loss": 0.8363,
+      "step": 176160
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6620104498301524e-05,
+      "loss": 0.8305,
+      "step": 176170
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.661934284887352e-05,
+      "loss": 0.9115,
+      "step": 176180
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6618581199445525e-05,
+      "loss": 0.9098,
+      "step": 176190
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6617819550017516e-05,
+      "loss": 0.8721,
+      "step": 176200
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661705790058952e-05,
+      "loss": 0.9054,
+      "step": 176210
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661629625116152e-05,
+      "loss": 0.8657,
+      "step": 176220
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6615534601733515e-05,
+      "loss": 0.8409,
+      "step": 176230
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661477295230551e-05,
+      "loss": 0.8773,
+      "step": 176240
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661401130287752e-05,
+      "loss": 0.8879,
+      "step": 176250
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661324965344951e-05,
+      "loss": 0.8926,
+      "step": 176260
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661248800402151e-05,
+      "loss": 0.8984,
+      "step": 176270
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661172635459351e-05,
+      "loss": 0.9278,
+      "step": 176280
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.661096470516551e-05,
+      "loss": 0.95,
+      "step": 176290
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6610203055737504e-05,
+      "loss": 0.8942,
+      "step": 176300
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.660944140630951e-05,
+      "loss": 0.9172,
+      "step": 176310
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6608679756881506e-05,
+      "loss": 0.7955,
+      "step": 176320
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6607918107453503e-05,
+      "loss": 0.9788,
+      "step": 176330
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.660715645802551e-05,
+      "loss": 0.858,
+      "step": 176340
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.66063948085975e-05,
+      "loss": 0.8982,
+      "step": 176350
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.66056331591695e-05,
+      "loss": 0.8525,
+      "step": 176360
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.66048715097415e-05,
+      "loss": 0.8454,
+      "step": 176370
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.66041098603135e-05,
+      "loss": 1.0099,
+      "step": 176380
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.6603348210885495e-05,
+      "loss": 0.8961,
+      "step": 176390
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.660258656145749e-05,
+      "loss": 0.9387,
+      "step": 176400
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.660182491202949e-05,
+      "loss": 0.8407,
+      "step": 176410
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6601063262601494e-05,
+      "loss": 0.8186,
+      "step": 176420
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6600301613173485e-05,
+      "loss": 0.905,
+      "step": 176430
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659953996374549e-05,
+      "loss": 0.8446,
+      "step": 176440
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659877831431749e-05,
+      "loss": 0.9512,
+      "step": 176450
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6598016664889484e-05,
+      "loss": 0.9043,
+      "step": 176460
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659725501546148e-05,
+      "loss": 0.9146,
+      "step": 176470
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6596493366033486e-05,
+      "loss": 0.8952,
+      "step": 176480
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659573171660548e-05,
+      "loss": 0.8298,
+      "step": 176490
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659497006717748e-05,
+      "loss": 0.8015,
+      "step": 176500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659420841774948e-05,
+      "loss": 0.8411,
+      "step": 176510
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6593446768321476e-05,
+      "loss": 0.9921,
+      "step": 176520
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659268511889348e-05,
+      "loss": 0.8918,
+      "step": 176530
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659192346946548e-05,
+      "loss": 0.8912,
+      "step": 176540
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6591161820037475e-05,
+      "loss": 0.8204,
+      "step": 176550
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.659040017060947e-05,
+      "loss": 0.8839,
+      "step": 176560
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6589638521181477e-05,
+      "loss": 0.9007,
+      "step": 176570
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.658887687175347e-05,
+      "loss": 0.8825,
+      "step": 176580
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.658811522232547e-05,
+      "loss": 0.9388,
+      "step": 176590
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.658735357289747e-05,
+      "loss": 0.915,
+      "step": 176600
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6586591923469466e-05,
+      "loss": 0.8815,
+      "step": 176610
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.6585830274041464e-05,
+      "loss": 0.9744,
+      "step": 176620
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.658506862461347e-05,
+      "loss": 0.8645,
+      "step": 176630
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.658430697518546e-05,
+      "loss": 0.8899,
+      "step": 176640
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.658354532575746e-05,
+      "loss": 0.8678,
+      "step": 176650
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.658278367632946e-05,
+      "loss": 0.8146,
+      "step": 176660
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.658202202690146e-05,
+      "loss": 0.895,
+      "step": 176670
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6581260377473455e-05,
+      "loss": 0.8435,
+      "step": 176680
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.658049872804546e-05,
+      "loss": 0.8985,
+      "step": 176690
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657973707861746e-05,
+      "loss": 0.8694,
+      "step": 176700
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6578975429189455e-05,
+      "loss": 0.8881,
+      "step": 176710
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657821377976145e-05,
+      "loss": 0.827,
+      "step": 176720
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657745213033345e-05,
+      "loss": 0.9263,
+      "step": 176730
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6576690480905454e-05,
+      "loss": 0.9113,
+      "step": 176740
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657592883147745e-05,
+      "loss": 0.9392,
+      "step": 176750
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657516718204945e-05,
+      "loss": 0.8852,
+      "step": 176760
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6574405532621446e-05,
+      "loss": 0.8328,
+      "step": 176770
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657364388319345e-05,
+      "loss": 0.724,
+      "step": 176780
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657288223376544e-05,
+      "loss": 0.8022,
+      "step": 176790
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6572120584337445e-05,
+      "loss": 1.0022,
+      "step": 176800
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657135893490944e-05,
+      "loss": 0.8874,
+      "step": 176810
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.657059728548144e-05,
+      "loss": 0.9071,
+      "step": 176820
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.656983563605344e-05,
+      "loss": 0.8227,
+      "step": 176830
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6569073986625435e-05,
+      "loss": 0.7936,
+      "step": 176840
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656831233719743e-05,
+      "loss": 0.872,
+      "step": 176850
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656755068776944e-05,
+      "loss": 0.8632,
+      "step": 176860
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.6566789038341434e-05,
+      "loss": 0.9281,
+      "step": 176870
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656602738891343e-05,
+      "loss": 0.8568,
+      "step": 176880
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656526573948543e-05,
+      "loss": 0.8285,
+      "step": 176890
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656450409005743e-05,
+      "loss": 0.8716,
+      "step": 176900
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656374244062943e-05,
+      "loss": 0.8765,
+      "step": 176910
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656298079120143e-05,
+      "loss": 0.81,
+      "step": 176920
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.6562219141773426e-05,
+      "loss": 0.9469,
+      "step": 176930
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.6561457492345423e-05,
+      "loss": 0.858,
+      "step": 176940
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.656069584291743e-05,
+      "loss": 0.9465,
+      "step": 176950
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655993419348942e-05,
+      "loss": 0.7864,
+      "step": 176960
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655917254406142e-05,
+      "loss": 0.9422,
+      "step": 176970
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655841089463342e-05,
+      "loss": 0.8827,
+      "step": 176980
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655764924520542e-05,
+      "loss": 0.7707,
+      "step": 176990
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.6556887595777415e-05,
+      "loss": 0.8749,
+      "step": 177000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655612594634942e-05,
+      "loss": 0.9433,
+      "step": 177010
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655536429692141e-05,
+      "loss": 0.9627,
+      "step": 177020
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.6554602647493414e-05,
+      "loss": 0.8867,
+      "step": 177030
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655384099806541e-05,
+      "loss": 0.9032,
+      "step": 177040
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.655307934863741e-05,
+      "loss": 0.8484,
+      "step": 177050
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.655231769920941e-05,
+      "loss": 0.8408,
+      "step": 177060
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.655155604978141e-05,
+      "loss": 0.8556,
+      "step": 177070
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.655079440035341e-05,
+      "loss": 0.815,
+      "step": 177080
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.6550032750925406e-05,
+      "loss": 0.9283,
+      "step": 177090
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.65492711014974e-05,
+      "loss": 0.902,
+      "step": 177100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.65485094520694e-05,
+      "loss": 0.8214,
+      "step": 177110
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.6547747802641405e-05,
+      "loss": 0.8807,
+      "step": 177120
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.65469861532134e-05,
+      "loss": 0.8833,
+      "step": 177130
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.65462245037854e-05,
+      "loss": 0.9357,
+      "step": 177140
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.65454628543574e-05,
+      "loss": 0.9537,
+      "step": 177150
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.65447012049294e-05,
+      "loss": 0.8737,
+      "step": 177160
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.654393955550139e-05,
+      "loss": 0.8993,
+      "step": 177170
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.6543177906073397e-05,
+      "loss": 0.9152,
+      "step": 177180
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.6542416256645394e-05,
+      "loss": 0.9113,
+      "step": 177190
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.654165460721739e-05,
+      "loss": 0.9431,
+      "step": 177200
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.654089295778939e-05,
+      "loss": 0.9073,
+      "step": 177210
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.654013130836139e-05,
+      "loss": 1.0363,
+      "step": 177220
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.6539369658933384e-05,
+      "loss": 0.9202,
+      "step": 177230
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.653860800950539e-05,
+      "loss": 0.7939,
+      "step": 177240
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.6537846360077386e-05,
+      "loss": 0.8395,
+      "step": 177250
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.653708471064938e-05,
+      "loss": 0.8884,
+      "step": 177260
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.653632306122138e-05,
+      "loss": 0.9206,
+      "step": 177270
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.6535561411793385e-05,
+      "loss": 0.905,
+      "step": 177280
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.653479976236538e-05,
+      "loss": 0.9328,
+      "step": 177290
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.653403811293738e-05,
+      "loss": 0.9818,
+      "step": 177300
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.653327646350938e-05,
+      "loss": 0.9102,
+      "step": 177310
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.6532514814081375e-05,
+      "loss": 0.9242,
+      "step": 177320
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.653175316465338e-05,
+      "loss": 0.8534,
+      "step": 177330
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.653099151522537e-05,
+      "loss": 0.8496,
+      "step": 177340
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.6530229865797374e-05,
+      "loss": 0.8396,
+      "step": 177350
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652946821636937e-05,
+      "loss": 0.8403,
+      "step": 177360
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652870656694137e-05,
+      "loss": 0.8573,
+      "step": 177370
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.6527944917513366e-05,
+      "loss": 0.9792,
+      "step": 177380
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652718326808537e-05,
+      "loss": 0.8099,
+      "step": 177390
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652642161865736e-05,
+      "loss": 0.8633,
+      "step": 177400
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.6525659969229365e-05,
+      "loss": 0.8373,
+      "step": 177410
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652489831980136e-05,
+      "loss": 0.9036,
+      "step": 177420
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652413667037336e-05,
+      "loss": 0.9243,
+      "step": 177430
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652337502094536e-05,
+      "loss": 0.8701,
+      "step": 177440
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652261337151736e-05,
+      "loss": 0.9126,
+      "step": 177450
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652185172208936e-05,
+      "loss": 0.9545,
+      "step": 177460
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.652109007266136e-05,
+      "loss": 0.9307,
+      "step": 177470
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.6520328423233354e-05,
+      "loss": 0.8051,
+      "step": 177480
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.651956677380535e-05,
+      "loss": 0.8836,
+      "step": 177490
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6518805124377356e-05,
+      "loss": 0.937,
+      "step": 177500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6518043474949354e-05,
+      "loss": 0.8053,
+      "step": 177510
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.651728182552135e-05,
+      "loss": 0.9809,
+      "step": 177520
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.651652017609335e-05,
+      "loss": 0.9825,
+      "step": 177530
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.651575852666535e-05,
+      "loss": 0.9483,
+      "step": 177540
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6514996877237344e-05,
+      "loss": 0.9224,
+      "step": 177550
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.651423522780935e-05,
+      "loss": 0.875,
+      "step": 177560
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6513473578381345e-05,
+      "loss": 0.8155,
+      "step": 177570
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.651271192895334e-05,
+      "loss": 0.8555,
+      "step": 177580
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.651195027952534e-05,
+      "loss": 0.8655,
+      "step": 177590
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6511188630097344e-05,
+      "loss": 0.8677,
+      "step": 177600
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6510426980669335e-05,
+      "loss": 0.8983,
+      "step": 177610
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.650966533124134e-05,
+      "loss": 0.8511,
+      "step": 177620
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.650890368181334e-05,
+      "loss": 0.8428,
+      "step": 177630
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6508142032385334e-05,
+      "loss": 0.8776,
+      "step": 177640
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.650738038295733e-05,
+      "loss": 0.8864,
+      "step": 177650
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6506618733529336e-05,
+      "loss": 0.828,
+      "step": 177660
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6505857084101333e-05,
+      "loss": 0.8875,
+      "step": 177670
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.650509543467333e-05,
+      "loss": 0.9226,
+      "step": 177680
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.650433378524533e-05,
+      "loss": 0.8205,
+      "step": 177690
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6503572135817326e-05,
+      "loss": 0.9484,
+      "step": 177700
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.650281048638933e-05,
+      "loss": 0.9267,
+      "step": 177710
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.650204883696133e-05,
+      "loss": 0.8818,
+      "step": 177720
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6501287187533325e-05,
+      "loss": 0.835,
+      "step": 177730
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.650052553810532e-05,
+      "loss": 0.8639,
+      "step": 177740
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649976388867733e-05,
+      "loss": 0.9668,
+      "step": 177750
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649900223924932e-05,
+      "loss": 0.8346,
+      "step": 177760
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649824058982132e-05,
+      "loss": 0.9251,
+      "step": 177770
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649747894039331e-05,
+      "loss": 0.7913,
+      "step": 177780
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6496717290965317e-05,
+      "loss": 0.8294,
+      "step": 177790
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6495955641537314e-05,
+      "loss": 0.8475,
+      "step": 177800
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649519399210931e-05,
+      "loss": 0.9512,
+      "step": 177810
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649443234268131e-05,
+      "loss": 0.9538,
+      "step": 177820
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649367069325331e-05,
+      "loss": 0.8583,
+      "step": 177830
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649290904382531e-05,
+      "loss": 0.9112,
+      "step": 177840
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.649214739439731e-05,
+      "loss": 0.8754,
+      "step": 177850
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6491385744969306e-05,
+      "loss": 0.8866,
+      "step": 177860
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.64906240955413e-05,
+      "loss": 0.7896,
+      "step": 177870
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.648986244611331e-05,
+      "loss": 0.8229,
+      "step": 177880
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6489100796685305e-05,
+      "loss": 0.7948,
+      "step": 177890
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.64883391472573e-05,
+      "loss": 0.899,
+      "step": 177900
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.64875774978293e-05,
+      "loss": 0.9176,
+      "step": 177910
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6486815848401304e-05,
+      "loss": 0.9578,
+      "step": 177920
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.6486054198973295e-05,
+      "loss": 0.875,
+      "step": 177930
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.64852925495453e-05,
+      "loss": 0.945,
+      "step": 177940
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6484530900117296e-05,
+      "loss": 0.8983,
+      "step": 177950
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6483769250689294e-05,
+      "loss": 0.9458,
+      "step": 177960
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.648300760126129e-05,
+      "loss": 0.8924,
+      "step": 177970
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6482245951833296e-05,
+      "loss": 0.8753,
+      "step": 177980
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6481484302405286e-05,
+      "loss": 0.7856,
+      "step": 177990
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.648072265297729e-05,
+      "loss": 0.8723,
+      "step": 178000
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647996100354929e-05,
+      "loss": 0.9352,
+      "step": 178010
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6479199354121285e-05,
+      "loss": 0.837,
+      "step": 178020
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647843770469328e-05,
+      "loss": 0.9842,
+      "step": 178030
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647767605526529e-05,
+      "loss": 0.9494,
+      "step": 178040
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6476914405837285e-05,
+      "loss": 0.8968,
+      "step": 178050
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647615275640928e-05,
+      "loss": 0.7921,
+      "step": 178060
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647539110698128e-05,
+      "loss": 0.8179,
+      "step": 178070
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647462945755328e-05,
+      "loss": 0.8852,
+      "step": 178080
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647386780812528e-05,
+      "loss": 0.9195,
+      "step": 178090
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647310615869728e-05,
+      "loss": 0.8521,
+      "step": 178100
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6472344509269276e-05,
+      "loss": 0.8413,
+      "step": 178110
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.6471582859841274e-05,
+      "loss": 0.8532,
+      "step": 178120
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647082121041328e-05,
+      "loss": 0.9019,
+      "step": 178130
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.647005956098527e-05,
+      "loss": 0.8304,
+      "step": 178140
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.646929791155727e-05,
+      "loss": 0.8284,
+      "step": 178150
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646853626212927e-05,
+      "loss": 0.852,
+      "step": 178160
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646777461270127e-05,
+      "loss": 0.8283,
+      "step": 178170
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6467012963273265e-05,
+      "loss": 0.8601,
+      "step": 178180
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646625131384527e-05,
+      "loss": 0.8671,
+      "step": 178190
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646548966441726e-05,
+      "loss": 0.8589,
+      "step": 178200
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6464728014989264e-05,
+      "loss": 0.889,
+      "step": 178210
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6463966365561255e-05,
+      "loss": 0.9049,
+      "step": 178220
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646320471613326e-05,
+      "loss": 0.9347,
+      "step": 178230
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646244306670526e-05,
+      "loss": 0.843,
+      "step": 178240
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6461681417277254e-05,
+      "loss": 0.9627,
+      "step": 178250
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.646091976784926e-05,
+      "loss": 0.9077,
+      "step": 178260
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6460158118421256e-05,
+      "loss": 0.9457,
+      "step": 178270
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6459396468993253e-05,
+      "loss": 0.9526,
+      "step": 178280
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.645863481956525e-05,
+      "loss": 0.8118,
+      "step": 178290
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6457873170137255e-05,
+      "loss": 0.8052,
+      "step": 178300
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6457111520709246e-05,
+      "loss": 0.8275,
+      "step": 178310
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.645634987128125e-05,
+      "loss": 0.8746,
+      "step": 178320
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.645558822185325e-05,
+      "loss": 0.8547,
+      "step": 178330
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6454826572425245e-05,
+      "loss": 0.9369,
+      "step": 178340
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.645406492299724e-05,
+      "loss": 0.8213,
+      "step": 178350
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.645330327356925e-05,
+      "loss": 0.8413,
+      "step": 178360
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.645254162414124e-05,
+      "loss": 0.8147,
+      "step": 178370
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.645177997471324e-05,
+      "loss": 0.8623,
+      "step": 178380
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.645101832528524e-05,
+      "loss": 0.8154,
+      "step": 178390
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6450256675857237e-05,
+      "loss": 0.9376,
+      "step": 178400
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6449495026429234e-05,
+      "loss": 0.9822,
+      "step": 178410
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644873337700124e-05,
+      "loss": 0.9407,
+      "step": 178420
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6447971727573236e-05,
+      "loss": 0.9287,
+      "step": 178430
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644721007814523e-05,
+      "loss": 0.7881,
+      "step": 178440
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644644842871723e-05,
+      "loss": 0.9256,
+      "step": 178450
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644568677928923e-05,
+      "loss": 0.806,
+      "step": 178460
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644492512986123e-05,
+      "loss": 0.9434,
+      "step": 178470
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644416348043323e-05,
+      "loss": 0.8071,
+      "step": 178480
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644340183100523e-05,
+      "loss": 0.8198,
+      "step": 178490
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6442640181577225e-05,
+      "loss": 0.9062,
+      "step": 178500
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644187853214923e-05,
+      "loss": 0.8858,
+      "step": 178510
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.644111688272122e-05,
+      "loss": 0.9085,
+      "step": 178520
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6440355233293224e-05,
+      "loss": 0.8573,
+      "step": 178530
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.643959358386522e-05,
+      "loss": 1.0451,
+      "step": 178540
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.643883193443722e-05,
+      "loss": 0.8354,
+      "step": 178550
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6438070285009216e-05,
+      "loss": 0.9137,
+      "step": 178560
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.643730863558122e-05,
+      "loss": 0.9117,
+      "step": 178570
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.643654698615321e-05,
+      "loss": 0.9665,
+      "step": 178580
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.6435785336725216e-05,
+      "loss": 0.9063,
+      "step": 178590
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.643502368729721e-05,
+      "loss": 0.9096,
+      "step": 178600
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.643426203786921e-05,
+      "loss": 0.8452,
+      "step": 178610
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.643350038844121e-05,
+      "loss": 0.8424,
+      "step": 178620
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.643273873901321e-05,
+      "loss": 0.8552,
+      "step": 178630
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.643197708958521e-05,
+      "loss": 0.8683,
+      "step": 178640
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.643121544015721e-05,
+      "loss": 0.825,
+      "step": 178650
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.6430453790729205e-05,
+      "loss": 0.8921,
+      "step": 178660
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.64296921413012e-05,
+      "loss": 1.001,
+      "step": 178670
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.6428930491873206e-05,
+      "loss": 0.9038,
+      "step": 178680
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.64281688424452e-05,
+      "loss": 0.9055,
+      "step": 178690
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.64274071930172e-05,
+      "loss": 1.0335,
+      "step": 178700
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.64266455435892e-05,
+      "loss": 0.8784,
+      "step": 178710
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.6425883894161196e-05,
+      "loss": 0.8728,
+      "step": 178720
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.6425122244733194e-05,
+      "loss": 0.8149,
+      "step": 178730
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.64243605953052e-05,
+      "loss": 0.8255,
+      "step": 178740
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.642359894587719e-05,
+      "loss": 0.8208,
+      "step": 178750
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.642283729644919e-05,
+      "loss": 0.9705,
+      "step": 178760
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.642207564702119e-05,
+      "loss": 0.8944,
+      "step": 178770
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.642131399759319e-05,
+      "loss": 0.8583,
+      "step": 178780
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.6420552348165185e-05,
+      "loss": 0.8627,
+      "step": 178790
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.641979069873719e-05,
+      "loss": 0.7567,
+      "step": 178800
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.641902904930918e-05,
+      "loss": 0.8782,
+      "step": 178810
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6418267399881184e-05,
+      "loss": 0.7925,
+      "step": 178820
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641750575045318e-05,
+      "loss": 0.8941,
+      "step": 178830
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641674410102518e-05,
+      "loss": 0.8996,
+      "step": 178840
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6415982451597184e-05,
+      "loss": 0.8739,
+      "step": 178850
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641522080216918e-05,
+      "loss": 0.7544,
+      "step": 178860
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641445915274118e-05,
+      "loss": 0.8654,
+      "step": 178870
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6413697503313176e-05,
+      "loss": 0.8458,
+      "step": 178880
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641293585388518e-05,
+      "loss": 0.9718,
+      "step": 178890
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641217420445717e-05,
+      "loss": 0.9153,
+      "step": 178900
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6411412555029175e-05,
+      "loss": 0.9126,
+      "step": 178910
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.641065090560117e-05,
+      "loss": 0.861,
+      "step": 178920
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640988925617317e-05,
+      "loss": 0.8938,
+      "step": 178930
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640912760674517e-05,
+      "loss": 0.8652,
+      "step": 178940
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640836595731717e-05,
+      "loss": 0.8481,
+      "step": 178950
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640760430788916e-05,
+      "loss": 0.9076,
+      "step": 178960
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640684265846117e-05,
+      "loss": 0.9788,
+      "step": 178970
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6406081009033164e-05,
+      "loss": 0.8347,
+      "step": 178980
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640531935960516e-05,
+      "loss": 0.8587,
+      "step": 178990
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640455771017716e-05,
+      "loss": 0.8077,
+      "step": 179000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6403796060749163e-05,
+      "loss": 0.8151,
+      "step": 179010
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.6403034411321154e-05,
+      "loss": 0.869,
+      "step": 179020
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.640227276189316e-05,
+      "loss": 0.9397,
+      "step": 179030
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6401511112465156e-05,
+      "loss": 0.78,
+      "step": 179040
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.640074946303715e-05,
+      "loss": 0.927,
+      "step": 179050
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639998781360916e-05,
+      "loss": 0.9375,
+      "step": 179060
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6399226164181155e-05,
+      "loss": 0.895,
+      "step": 179070
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639846451475315e-05,
+      "loss": 0.8792,
+      "step": 179080
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639770286532515e-05,
+      "loss": 0.8258,
+      "step": 179090
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6396941215897154e-05,
+      "loss": 0.8534,
+      "step": 179100
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6396179566469145e-05,
+      "loss": 0.882,
+      "step": 179110
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639541791704115e-05,
+      "loss": 0.8539,
+      "step": 179120
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6394656267613147e-05,
+      "loss": 0.838,
+      "step": 179130
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6393894618185144e-05,
+      "loss": 0.9067,
+      "step": 179140
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639313296875714e-05,
+      "loss": 0.9529,
+      "step": 179150
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639237131932914e-05,
+      "loss": 0.8973,
+      "step": 179160
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6391609669901136e-05,
+      "loss": 0.9056,
+      "step": 179170
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639084802047314e-05,
+      "loss": 0.9212,
+      "step": 179180
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.639008637104513e-05,
+      "loss": 0.9207,
+      "step": 179190
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6389324721617136e-05,
+      "loss": 0.9252,
+      "step": 179200
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.638856307218913e-05,
+      "loss": 0.797,
+      "step": 179210
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.638780142276113e-05,
+      "loss": 0.8119,
+      "step": 179220
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.6387039773333135e-05,
+      "loss": 0.8987,
+      "step": 179230
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.638627812390513e-05,
+      "loss": 0.8187,
+      "step": 179240
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.638551647447713e-05,
+      "loss": 0.8107,
+      "step": 179250
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.638475482504913e-05,
+      "loss": 0.9051,
+      "step": 179260
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.638399317562113e-05,
+      "loss": 0.858,
+      "step": 179270
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.638323152619312e-05,
+      "loss": 0.8408,
+      "step": 179280
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6382469876765126e-05,
+      "loss": 0.8459,
+      "step": 179290
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6381708227337124e-05,
+      "loss": 0.9174,
+      "step": 179300
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.638094657790912e-05,
+      "loss": 0.9547,
+      "step": 179310
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.638018492848112e-05,
+      "loss": 0.972,
+      "step": 179320
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637942327905312e-05,
+      "loss": 0.813,
+      "step": 179330
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6378661629625114e-05,
+      "loss": 0.8203,
+      "step": 179340
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637789998019712e-05,
+      "loss": 0.9067,
+      "step": 179350
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6377138330769115e-05,
+      "loss": 0.9255,
+      "step": 179360
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637637668134111e-05,
+      "loss": 0.9023,
+      "step": 179370
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637561503191311e-05,
+      "loss": 0.8343,
+      "step": 179380
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6374853382485115e-05,
+      "loss": 0.8279,
+      "step": 179390
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6374091733057105e-05,
+      "loss": 0.9504,
+      "step": 179400
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637333008362911e-05,
+      "loss": 0.887,
+      "step": 179410
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637256843420111e-05,
+      "loss": 0.8843,
+      "step": 179420
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6371806784773104e-05,
+      "loss": 0.8247,
+      "step": 179430
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.637104513534511e-05,
+      "loss": 0.8531,
+      "step": 179440
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6370283485917106e-05,
+      "loss": 0.8975,
+      "step": 179450
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.6369521836489104e-05,
+      "loss": 0.8395,
+      "step": 179460
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.63687601870611e-05,
+      "loss": 0.8757,
+      "step": 179470
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.6367998537633105e-05,
+      "loss": 0.9098,
+      "step": 179480
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.6367236888205096e-05,
+      "loss": 0.9811,
+      "step": 179490
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.63664752387771e-05,
+      "loss": 0.8509,
+      "step": 179500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.63657135893491e-05,
+      "loss": 0.8636,
+      "step": 179510
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.6364951939921095e-05,
+      "loss": 0.9513,
+      "step": 179520
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.636419029049309e-05,
+      "loss": 0.9285,
+      "step": 179530
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.63634286410651e-05,
+      "loss": 0.9477,
+      "step": 179540
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.636266699163709e-05,
+      "loss": 0.8337,
+      "step": 179550
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.636190534220909e-05,
+      "loss": 0.9203,
+      "step": 179560
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.636114369278109e-05,
+      "loss": 0.887,
+      "step": 179570
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.636038204335309e-05,
+      "loss": 0.8995,
+      "step": 179580
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.6359620393925084e-05,
+      "loss": 0.7432,
+      "step": 179590
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635885874449709e-05,
+      "loss": 0.8739,
+      "step": 179600
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635809709506908e-05,
+      "loss": 0.8052,
+      "step": 179610
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.6357335445641083e-05,
+      "loss": 0.8842,
+      "step": 179620
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635657379621308e-05,
+      "loss": 0.9725,
+      "step": 179630
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635581214678508e-05,
+      "loss": 0.8215,
+      "step": 179640
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635505049735708e-05,
+      "loss": 0.8374,
+      "step": 179650
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635428884792907e-05,
+      "loss": 0.7898,
+      "step": 179660
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.635352719850108e-05,
+      "loss": 0.8516,
+      "step": 179670
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.6352765549073075e-05,
+      "loss": 0.9442,
+      "step": 179680
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.635200389964507e-05,
+      "loss": 0.7806,
+      "step": 179690
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.635124225021707e-05,
+      "loss": 0.961,
+      "step": 179700
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6350480600789074e-05,
+      "loss": 0.92,
+      "step": 179710
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6349718951361065e-05,
+      "loss": 0.9417,
+      "step": 179720
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.634895730193307e-05,
+      "loss": 0.9461,
+      "step": 179730
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6348195652505067e-05,
+      "loss": 0.8803,
+      "step": 179740
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6347434003077064e-05,
+      "loss": 0.8501,
+      "step": 179750
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.634667235364906e-05,
+      "loss": 0.8847,
+      "step": 179760
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6345910704221066e-05,
+      "loss": 0.9264,
+      "step": 179770
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6345149054793056e-05,
+      "loss": 0.9544,
+      "step": 179780
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.634438740536506e-05,
+      "loss": 0.878,
+      "step": 179790
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.634362575593706e-05,
+      "loss": 0.8594,
+      "step": 179800
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6342864106509056e-05,
+      "loss": 0.8614,
+      "step": 179810
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.634210245708105e-05,
+      "loss": 0.8419,
+      "step": 179820
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.634134080765306e-05,
+      "loss": 0.9368,
+      "step": 179830
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6340579158225055e-05,
+      "loss": 0.973,
+      "step": 179840
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.633981750879705e-05,
+      "loss": 0.7947,
+      "step": 179850
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6339055859369056e-05,
+      "loss": 0.927,
+      "step": 179860
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.633829420994105e-05,
+      "loss": 0.9028,
+      "step": 179870
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.633753256051305e-05,
+      "loss": 0.9405,
+      "step": 179880
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.633677091108505e-05,
+      "loss": 0.8943,
+      "step": 179890
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6336009261657046e-05,
+      "loss": 0.9928,
+      "step": 179900
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6335247612229044e-05,
+      "loss": 0.9496,
+      "step": 179910
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.633448596280105e-05,
+      "loss": 0.7869,
+      "step": 179920
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.633372431337304e-05,
+      "loss": 0.9274,
+      "step": 179930
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.633296266394504e-05,
+      "loss": 0.7905,
+      "step": 179940
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.633220101451704e-05,
+      "loss": 0.8493,
+      "step": 179950
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.633143936508904e-05,
+      "loss": 0.9099,
+      "step": 179960
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6330677715661035e-05,
+      "loss": 0.8235,
+      "step": 179970
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632991606623304e-05,
+      "loss": 0.8113,
+      "step": 179980
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632915441680503e-05,
+      "loss": 0.8683,
+      "step": 179990
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6328392767377035e-05,
+      "loss": 0.8963,
+      "step": 180000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632763111794903e-05,
+      "loss": 0.9912,
+      "step": 180010
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632686946852103e-05,
+      "loss": 0.9327,
+      "step": 180020
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6326107819093034e-05,
+      "loss": 0.9742,
+      "step": 180030
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632534616966503e-05,
+      "loss": 0.8425,
+      "step": 180040
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632458452023703e-05,
+      "loss": 0.8728,
+      "step": 180050
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6323822870809026e-05,
+      "loss": 0.8471,
+      "step": 180060
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6323061221381024e-05,
+      "loss": 0.9105,
+      "step": 180070
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632229957195302e-05,
+      "loss": 0.9715,
+      "step": 180080
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6321537922525025e-05,
+      "loss": 0.9258,
+      "step": 180090
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6320776273097016e-05,
+      "loss": 0.9327,
+      "step": 180100
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.632001462366902e-05,
+      "loss": 0.8387,
+      "step": 180110
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.631925297424102e-05,
+      "loss": 0.9054,
+      "step": 180120
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6318491324813015e-05,
+      "loss": 0.7755,
+      "step": 180130
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631772967538501e-05,
+      "loss": 0.8493,
+      "step": 180140
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631696802595702e-05,
+      "loss": 0.8358,
+      "step": 180150
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631620637652901e-05,
+      "loss": 0.8543,
+      "step": 180160
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631544472710101e-05,
+      "loss": 0.8231,
+      "step": 180170
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631468307767301e-05,
+      "loss": 0.9327,
+      "step": 180180
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631392142824501e-05,
+      "loss": 0.9045,
+      "step": 180190
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6313159778817004e-05,
+      "loss": 0.8891,
+      "step": 180200
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631239812938901e-05,
+      "loss": 0.8527,
+      "step": 180210
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6311636479961006e-05,
+      "loss": 0.907,
+      "step": 180220
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6310874830533003e-05,
+      "loss": 0.8607,
+      "step": 180230
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.631011318110501e-05,
+      "loss": 0.7913,
+      "step": 180240
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6309351531677e-05,
+      "loss": 0.8771,
+      "step": 180250
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6308589882249e-05,
+      "loss": 0.9722,
+      "step": 180260
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6307828232821e-05,
+      "loss": 0.9013,
+      "step": 180270
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6307066583393e-05,
+      "loss": 0.8457,
+      "step": 180280
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6306304933964995e-05,
+      "loss": 0.7794,
+      "step": 180290
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6305543284537e-05,
+      "loss": 0.8938,
+      "step": 180300
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.630478163510899e-05,
+      "loss": 0.8484,
+      "step": 180310
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6304019985680994e-05,
+      "loss": 0.9457,
+      "step": 180320
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.630325833625299e-05,
+      "loss": 0.9616,
+      "step": 180330
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.630249668682499e-05,
+      "loss": 0.8782,
+      "step": 180340
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.6301735037396987e-05,
+      "loss": 0.8308,
+      "step": 180350
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.630097338796899e-05,
+      "loss": 0.85,
+      "step": 180360
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.630021173854098e-05,
+      "loss": 0.9263,
+      "step": 180370
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.6299450089112986e-05,
+      "loss": 0.8875,
+      "step": 180380
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629868843968498e-05,
+      "loss": 0.8293,
+      "step": 180390
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629792679025698e-05,
+      "loss": 0.8629,
+      "step": 180400
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629716514082898e-05,
+      "loss": 0.7766,
+      "step": 180410
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629640349140098e-05,
+      "loss": 0.8518,
+      "step": 180420
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629564184197298e-05,
+      "loss": 0.8802,
+      "step": 180430
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629488019254498e-05,
+      "loss": 0.9184,
+      "step": 180440
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629411854311698e-05,
+      "loss": 0.8647,
+      "step": 180450
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629335689368897e-05,
+      "loss": 0.8728,
+      "step": 180460
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.6292595244260976e-05,
+      "loss": 0.8291,
+      "step": 180470
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.6291833594832974e-05,
+      "loss": 1.0096,
+      "step": 180480
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629107194540497e-05,
+      "loss": 0.8922,
+      "step": 180490
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.629031029597697e-05,
+      "loss": 0.8356,
+      "step": 180500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.628954864654897e-05,
+      "loss": 0.85,
+      "step": 180510
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.6288786997120964e-05,
+      "loss": 0.8571,
+      "step": 180520
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.628802534769297e-05,
+      "loss": 0.8635,
+      "step": 180530
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.628726369826496e-05,
+      "loss": 0.9806,
+      "step": 180540
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.628650204883696e-05,
+      "loss": 0.9269,
+      "step": 180550
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.628574039940896e-05,
+      "loss": 0.9351,
+      "step": 180560
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.628497874998096e-05,
+      "loss": 0.9245,
+      "step": 180570
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6284217100552955e-05,
+      "loss": 0.8601,
+      "step": 180580
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.628345545112496e-05,
+      "loss": 0.8998,
+      "step": 180590
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.628269380169696e-05,
+      "loss": 0.859,
+      "step": 180600
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6281932152268955e-05,
+      "loss": 0.9272,
+      "step": 180610
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.628117050284095e-05,
+      "loss": 0.8698,
+      "step": 180620
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.628040885341295e-05,
+      "loss": 0.8648,
+      "step": 180630
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6279647203984954e-05,
+      "loss": 0.8495,
+      "step": 180640
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627888555455695e-05,
+      "loss": 0.8452,
+      "step": 180650
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627812390512895e-05,
+      "loss": 0.8486,
+      "step": 180660
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6277362255700946e-05,
+      "loss": 0.8664,
+      "step": 180670
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627660060627295e-05,
+      "loss": 0.8908,
+      "step": 180680
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627583895684494e-05,
+      "loss": 0.9079,
+      "step": 180690
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6275077307416945e-05,
+      "loss": 1.0135,
+      "step": 180700
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627431565798894e-05,
+      "loss": 0.9726,
+      "step": 180710
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627355400856094e-05,
+      "loss": 0.9488,
+      "step": 180720
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627279235913294e-05,
+      "loss": 0.8439,
+      "step": 180730
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627203070970494e-05,
+      "loss": 0.8605,
+      "step": 180740
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627126906027693e-05,
+      "loss": 0.8952,
+      "step": 180750
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.627050741084894e-05,
+      "loss": 0.8676,
+      "step": 180760
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6269745761420934e-05,
+      "loss": 1.0016,
+      "step": 180770
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.626898411199293e-05,
+      "loss": 0.8866,
+      "step": 180780
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626822246256493e-05,
+      "loss": 0.9646,
+      "step": 180790
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6267460813136934e-05,
+      "loss": 0.958,
+      "step": 180800
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626669916370893e-05,
+      "loss": 0.8908,
+      "step": 180810
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626593751428093e-05,
+      "loss": 0.937,
+      "step": 180820
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626517586485293e-05,
+      "loss": 0.9231,
+      "step": 180830
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6264414215424923e-05,
+      "loss": 0.9399,
+      "step": 180840
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626365256599693e-05,
+      "loss": 0.9554,
+      "step": 180850
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6262890916568925e-05,
+      "loss": 0.8819,
+      "step": 180860
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626212926714092e-05,
+      "loss": 0.9074,
+      "step": 180870
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.626136761771292e-05,
+      "loss": 0.8949,
+      "step": 180880
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6260605968284924e-05,
+      "loss": 0.926,
+      "step": 180890
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6259844318856915e-05,
+      "loss": 0.8839,
+      "step": 180900
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.625908266942892e-05,
+      "loss": 0.9562,
+      "step": 180910
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.625832102000092e-05,
+      "loss": 0.9889,
+      "step": 180920
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6257559370572914e-05,
+      "loss": 0.807,
+      "step": 180930
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.625679772114491e-05,
+      "loss": 0.9369,
+      "step": 180940
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6256036071716916e-05,
+      "loss": 0.852,
+      "step": 180950
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6255274422288907e-05,
+      "loss": 0.8123,
+      "step": 180960
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.625451277286091e-05,
+      "loss": 0.8447,
+      "step": 180970
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.625375112343291e-05,
+      "loss": 0.8164,
+      "step": 180980
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.6252989474004906e-05,
+      "loss": 0.8764,
+      "step": 180990
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.62522278245769e-05,
+      "loss": 0.9353,
+      "step": 181000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.62514661751489e-05,
+      "loss": 0.9615,
+      "step": 181010
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6250704525720905e-05,
+      "loss": 0.9104,
+      "step": 181020
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.62499428762929e-05,
+      "loss": 0.878,
+      "step": 181030
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.62491812268649e-05,
+      "loss": 0.8428,
+      "step": 181040
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.62484195774369e-05,
+      "loss": 0.8485,
+      "step": 181050
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.62476579280089e-05,
+      "loss": 0.9147,
+      "step": 181060
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.624689627858089e-05,
+      "loss": 0.9136,
+      "step": 181070
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6246134629152897e-05,
+      "loss": 0.8773,
+      "step": 181080
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6245372979724894e-05,
+      "loss": 0.9679,
+      "step": 181090
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.624461133029689e-05,
+      "loss": 1.0295,
+      "step": 181100
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.624384968086889e-05,
+      "loss": 0.7915,
+      "step": 181110
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.624308803144089e-05,
+      "loss": 0.9717,
+      "step": 181120
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6242326382012884e-05,
+      "loss": 0.887,
+      "step": 181130
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.624156473258489e-05,
+      "loss": 0.7932,
+      "step": 181140
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6240803083156886e-05,
+      "loss": 0.8864,
+      "step": 181150
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.624004143372888e-05,
+      "loss": 0.8818,
+      "step": 181160
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.623927978430088e-05,
+      "loss": 0.9805,
+      "step": 181170
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6238518134872885e-05,
+      "loss": 0.9704,
+      "step": 181180
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.623775648544488e-05,
+      "loss": 0.9588,
+      "step": 181190
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.623699483601688e-05,
+      "loss": 0.8661,
+      "step": 181200
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.623623318658888e-05,
+      "loss": 0.9229,
+      "step": 181210
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.6235471537160875e-05,
+      "loss": 0.8579,
+      "step": 181220
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.623470988773288e-05,
+      "loss": 0.865,
+      "step": 181230
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6233948238304876e-05,
+      "loss": 0.8626,
+      "step": 181240
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6233186588876874e-05,
+      "loss": 0.8968,
+      "step": 181250
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.623242493944887e-05,
+      "loss": 0.9118,
+      "step": 181260
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6231663290020875e-05,
+      "loss": 0.8165,
+      "step": 181270
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6230901640592866e-05,
+      "loss": 0.8404,
+      "step": 181280
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.623013999116487e-05,
+      "loss": 0.8602,
+      "step": 181290
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622937834173687e-05,
+      "loss": 0.9576,
+      "step": 181300
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6228616692308865e-05,
+      "loss": 0.9448,
+      "step": 181310
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622785504288086e-05,
+      "loss": 0.8597,
+      "step": 181320
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622709339345287e-05,
+      "loss": 0.9076,
+      "step": 181330
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622633174402486e-05,
+      "loss": 0.8833,
+      "step": 181340
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622557009459686e-05,
+      "loss": 0.8102,
+      "step": 181350
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622480844516886e-05,
+      "loss": 0.8875,
+      "step": 181360
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622404679574086e-05,
+      "loss": 0.8794,
+      "step": 181370
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6223285146312854e-05,
+      "loss": 0.964,
+      "step": 181380
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622252349688486e-05,
+      "loss": 0.8701,
+      "step": 181390
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6221761847456856e-05,
+      "loss": 0.8787,
+      "step": 181400
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6221000198028854e-05,
+      "loss": 1.0008,
+      "step": 181410
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.622023854860085e-05,
+      "loss": 0.8998,
+      "step": 181420
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.621947689917285e-05,
+      "loss": 0.9005,
+      "step": 181430
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621871524974485e-05,
+      "loss": 0.8464,
+      "step": 181440
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6217953600316843e-05,
+      "loss": 0.9722,
+      "step": 181450
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621719195088885e-05,
+      "loss": 0.9606,
+      "step": 181460
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6216430301460845e-05,
+      "loss": 0.8316,
+      "step": 181470
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621566865203284e-05,
+      "loss": 0.8367,
+      "step": 181480
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621490700260484e-05,
+      "loss": 0.8314,
+      "step": 181490
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6214145353176844e-05,
+      "loss": 0.9289,
+      "step": 181500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6213383703748835e-05,
+      "loss": 0.8748,
+      "step": 181510
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621262205432084e-05,
+      "loss": 0.8786,
+      "step": 181520
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621186040489284e-05,
+      "loss": 0.8226,
+      "step": 181530
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6211098755464834e-05,
+      "loss": 0.852,
+      "step": 181540
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.621033710603683e-05,
+      "loss": 0.8937,
+      "step": 181550
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6209575456608836e-05,
+      "loss": 0.892,
+      "step": 181560
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620881380718083e-05,
+      "loss": 0.9671,
+      "step": 181570
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620805215775283e-05,
+      "loss": 0.9166,
+      "step": 181580
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620729050832483e-05,
+      "loss": 0.8314,
+      "step": 181590
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6206528858896826e-05,
+      "loss": 0.9127,
+      "step": 181600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620576720946883e-05,
+      "loss": 0.8026,
+      "step": 181610
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620500556004083e-05,
+      "loss": 0.9107,
+      "step": 181620
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.6204243910612825e-05,
+      "loss": 0.8558,
+      "step": 181630
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620348226118482e-05,
+      "loss": 0.9114,
+      "step": 181640
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.620272061175683e-05,
+      "loss": 0.804,
+      "step": 181650
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.620195896232882e-05,
+      "loss": 0.9477,
+      "step": 181660
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.620119731290082e-05,
+      "loss": 0.8098,
+      "step": 181670
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.620043566347282e-05,
+      "loss": 0.8679,
+      "step": 181680
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.6199674014044817e-05,
+      "loss": 0.9329,
+      "step": 181690
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.6198912364616814e-05,
+      "loss": 0.8662,
+      "step": 181700
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619815071518882e-05,
+      "loss": 0.8675,
+      "step": 181710
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619738906576081e-05,
+      "loss": 0.9143,
+      "step": 181720
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619662741633281e-05,
+      "loss": 0.8167,
+      "step": 181730
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619586576690481e-05,
+      "loss": 0.8932,
+      "step": 181740
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619510411747681e-05,
+      "loss": 0.9242,
+      "step": 181750
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.6194342468048806e-05,
+      "loss": 0.8974,
+      "step": 181760
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619358081862081e-05,
+      "loss": 0.8746,
+      "step": 181770
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.619281916919281e-05,
+      "loss": 0.8565,
+      "step": 181780
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.6192057519764805e-05,
+      "loss": 0.923,
+      "step": 181790
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.61912958703368e-05,
+      "loss": 0.9277,
+      "step": 181800
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.61905342209088e-05,
+      "loss": 0.8792,
+      "step": 181810
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.6189772571480804e-05,
+      "loss": 0.8658,
+      "step": 181820
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.61890109220528e-05,
+      "loss": 0.982,
+      "step": 181830
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.61882492726248e-05,
+      "loss": 0.8951,
+      "step": 181840
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.6187487623196796e-05,
+      "loss": 0.959,
+      "step": 181850
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.61867259737688e-05,
+      "loss": 0.9142,
+      "step": 181860
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.618596432434079e-05,
+      "loss": 0.9003,
+      "step": 181870
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6185202674912795e-05,
+      "loss": 0.9595,
+      "step": 181880
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.618444102548479e-05,
+      "loss": 0.8028,
+      "step": 181890
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.618367937605679e-05,
+      "loss": 0.8474,
+      "step": 181900
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.618291772662879e-05,
+      "loss": 0.9117,
+      "step": 181910
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6182156077200785e-05,
+      "loss": 0.9974,
+      "step": 181920
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.618139442777278e-05,
+      "loss": 0.8216,
+      "step": 181930
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.618063277834479e-05,
+      "loss": 0.932,
+      "step": 181940
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6179871128916785e-05,
+      "loss": 0.8988,
+      "step": 181950
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617910947948878e-05,
+      "loss": 0.8857,
+      "step": 181960
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617834783006078e-05,
+      "loss": 0.821,
+      "step": 181970
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617758618063278e-05,
+      "loss": 0.8597,
+      "step": 181980
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617682453120478e-05,
+      "loss": 0.9923,
+      "step": 181990
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617606288177678e-05,
+      "loss": 0.8764,
+      "step": 182000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6175301232348776e-05,
+      "loss": 0.9191,
+      "step": 182010
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6174539582920774e-05,
+      "loss": 0.7847,
+      "step": 182020
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617377793349278e-05,
+      "loss": 0.8548,
+      "step": 182030
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617301628406477e-05,
+      "loss": 0.8926,
+      "step": 182040
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617225463463677e-05,
+      "loss": 0.9228,
+      "step": 182050
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617149298520877e-05,
+      "loss": 0.9477,
+      "step": 182060
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.617073133578077e-05,
+      "loss": 0.7579,
+      "step": 182070
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6169969686352765e-05,
+      "loss": 0.9005,
+      "step": 182080
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.616920803692477e-05,
+      "loss": 0.8191,
+      "step": 182090
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616844638749676e-05,
+      "loss": 0.9076,
+      "step": 182100
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6167684738068764e-05,
+      "loss": 0.8871,
+      "step": 182110
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616692308864076e-05,
+      "loss": 0.7608,
+      "step": 182120
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616616143921276e-05,
+      "loss": 0.7554,
+      "step": 182130
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616539978978476e-05,
+      "loss": 0.9096,
+      "step": 182140
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616463814035676e-05,
+      "loss": 0.8236,
+      "step": 182150
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616387649092876e-05,
+      "loss": 0.8951,
+      "step": 182160
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6163114841500756e-05,
+      "loss": 0.8185,
+      "step": 182170
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616235319207275e-05,
+      "loss": 0.8563,
+      "step": 182180
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616159154264475e-05,
+      "loss": 0.8629,
+      "step": 182190
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6160829893216755e-05,
+      "loss": 0.9916,
+      "step": 182200
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.616006824378875e-05,
+      "loss": 0.8302,
+      "step": 182210
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615930659436075e-05,
+      "loss": 0.8855,
+      "step": 182220
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615854494493275e-05,
+      "loss": 0.71,
+      "step": 182230
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615778329550475e-05,
+      "loss": 0.8369,
+      "step": 182240
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615702164607674e-05,
+      "loss": 0.8397,
+      "step": 182250
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615625999664875e-05,
+      "loss": 0.9923,
+      "step": 182260
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6155498347220744e-05,
+      "loss": 0.929,
+      "step": 182270
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615473669779274e-05,
+      "loss": 0.9056,
+      "step": 182280
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615397504836474e-05,
+      "loss": 0.8012,
+      "step": 182290
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.615321339893674e-05,
+      "loss": 1.0282,
+      "step": 182300
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6152451749508734e-05,
+      "loss": 0.8275,
+      "step": 182310
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.615169010008074e-05,
+      "loss": 0.8946,
+      "step": 182320
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6150928450652736e-05,
+      "loss": 0.891,
+      "step": 182330
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.615016680122473e-05,
+      "loss": 0.9551,
+      "step": 182340
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614940515179673e-05,
+      "loss": 0.9875,
+      "step": 182350
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6148643502368735e-05,
+      "loss": 0.9334,
+      "step": 182360
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614788185294073e-05,
+      "loss": 0.7855,
+      "step": 182370
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614712020351273e-05,
+      "loss": 0.9097,
+      "step": 182380
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614635855408473e-05,
+      "loss": 0.7806,
+      "step": 182390
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6145596904656725e-05,
+      "loss": 0.856,
+      "step": 182400
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614483525522873e-05,
+      "loss": 0.9069,
+      "step": 182410
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614407360580072e-05,
+      "loss": 0.8929,
+      "step": 182420
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6143311956372724e-05,
+      "loss": 0.8341,
+      "step": 182430
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614255030694472e-05,
+      "loss": 1.0043,
+      "step": 182440
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614178865751672e-05,
+      "loss": 0.8329,
+      "step": 182450
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6141027008088716e-05,
+      "loss": 0.8554,
+      "step": 182460
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.614026535866072e-05,
+      "loss": 0.8306,
+      "step": 182470
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.613950370923271e-05,
+      "loss": 0.9529,
+      "step": 182480
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6138742059804715e-05,
+      "loss": 0.8895,
+      "step": 182490
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.613798041037671e-05,
+      "loss": 0.8101,
+      "step": 182500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.613721876094871e-05,
+      "loss": 0.8762,
+      "step": 182510
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.613645711152071e-05,
+      "loss": 0.9534,
+      "step": 182520
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.613569546209271e-05,
+      "loss": 0.9162,
+      "step": 182530
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.613493381266471e-05,
+      "loss": 0.8409,
+      "step": 182540
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.613417216323671e-05,
+      "loss": 0.8749,
+      "step": 182550
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6133410513808705e-05,
+      "loss": 1.0422,
+      "step": 182560
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.61326488643807e-05,
+      "loss": 0.9697,
+      "step": 182570
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6131887214952706e-05,
+      "loss": 0.8798,
+      "step": 182580
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6131125565524704e-05,
+      "loss": 0.8952,
+      "step": 182590
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.61303639160967e-05,
+      "loss": 0.8283,
+      "step": 182600
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.61296022666687e-05,
+      "loss": 0.8852,
+      "step": 182610
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.61288406172407e-05,
+      "loss": 0.9321,
+      "step": 182620
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6128078967812694e-05,
+      "loss": 0.8758,
+      "step": 182630
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.61273173183847e-05,
+      "loss": 0.8607,
+      "step": 182640
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6126555668956695e-05,
+      "loss": 0.9325,
+      "step": 182650
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.612579401952869e-05,
+      "loss": 0.7966,
+      "step": 182660
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.612503237010069e-05,
+      "loss": 0.9135,
+      "step": 182670
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6124270720672694e-05,
+      "loss": 0.8674,
+      "step": 182680
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6123509071244685e-05,
+      "loss": 0.8748,
+      "step": 182690
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.612274742181669e-05,
+      "loss": 0.8549,
+      "step": 182700
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.612198577238869e-05,
+      "loss": 0.8316,
+      "step": 182710
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6121224122960684e-05,
+      "loss": 0.8282,
+      "step": 182720
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.612046247353268e-05,
+      "loss": 0.8994,
+      "step": 182730
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6119700824104686e-05,
+      "loss": 0.8916,
+      "step": 182740
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.6118939174676684e-05,
+      "loss": 0.8849,
+      "step": 182750
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611817752524868e-05,
+      "loss": 0.8733,
+      "step": 182760
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611741587582068e-05,
+      "loss": 0.9062,
+      "step": 182770
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.6116654226392676e-05,
+      "loss": 0.9182,
+      "step": 182780
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611589257696468e-05,
+      "loss": 0.7721,
+      "step": 182790
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611513092753668e-05,
+      "loss": 0.8308,
+      "step": 182800
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.6114369278108675e-05,
+      "loss": 0.8815,
+      "step": 182810
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611360762868067e-05,
+      "loss": 0.8294,
+      "step": 182820
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611284597925267e-05,
+      "loss": 0.8807,
+      "step": 182830
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611208432982467e-05,
+      "loss": 0.7805,
+      "step": 182840
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611132268039667e-05,
+      "loss": 0.8156,
+      "step": 182850
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.611056103096866e-05,
+      "loss": 0.8337,
+      "step": 182860
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610979938154067e-05,
+      "loss": 0.8715,
+      "step": 182870
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.6109037732112664e-05,
+      "loss": 0.8469,
+      "step": 182880
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610827608268466e-05,
+      "loss": 0.9493,
+      "step": 182890
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610751443325666e-05,
+      "loss": 0.9473,
+      "step": 182900
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610675278382866e-05,
+      "loss": 0.8234,
+      "step": 182910
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610599113440066e-05,
+      "loss": 0.916,
+      "step": 182920
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610522948497266e-05,
+      "loss": 0.8754,
+      "step": 182930
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.6104467835544656e-05,
+      "loss": 0.9333,
+      "step": 182940
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610370618611665e-05,
+      "loss": 0.8223,
+      "step": 182950
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.610294453668866e-05,
+      "loss": 0.9457,
+      "step": 182960
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.6102182887260655e-05,
+      "loss": 0.9056,
+      "step": 182970
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.610142123783265e-05,
+      "loss": 0.8801,
+      "step": 182980
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.610065958840465e-05,
+      "loss": 0.8501,
+      "step": 182990
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6099897938976654e-05,
+      "loss": 0.9542,
+      "step": 183000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6099136289548645e-05,
+      "loss": 0.8977,
+      "step": 183010
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.609837464012065e-05,
+      "loss": 0.9186,
+      "step": 183020
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6097612990692646e-05,
+      "loss": 0.9426,
+      "step": 183030
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6096851341264644e-05,
+      "loss": 0.7873,
+      "step": 183040
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.609608969183664e-05,
+      "loss": 0.8586,
+      "step": 183050
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6095328042408646e-05,
+      "loss": 0.8597,
+      "step": 183060
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6094566392980636e-05,
+      "loss": 0.7866,
+      "step": 183070
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.609380474355264e-05,
+      "loss": 0.856,
+      "step": 183080
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.609304309412464e-05,
+      "loss": 0.9305,
+      "step": 183090
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6092281444696636e-05,
+      "loss": 0.998,
+      "step": 183100
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.609151979526863e-05,
+      "loss": 0.9596,
+      "step": 183110
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.609075814584064e-05,
+      "loss": 0.8638,
+      "step": 183120
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6089996496412635e-05,
+      "loss": 0.9324,
+      "step": 183130
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.608923484698463e-05,
+      "loss": 0.9412,
+      "step": 183140
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.608847319755663e-05,
+      "loss": 0.8455,
+      "step": 183150
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.608771154812863e-05,
+      "loss": 0.9205,
+      "step": 183160
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.608694989870063e-05,
+      "loss": 0.9687,
+      "step": 183170
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.608618824927263e-05,
+      "loss": 0.8834,
+      "step": 183180
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.6085426599844626e-05,
+      "loss": 0.8115,
+      "step": 183190
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6084664950416624e-05,
+      "loss": 0.8788,
+      "step": 183200
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.608390330098863e-05,
+      "loss": 0.9601,
+      "step": 183210
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.608314165156062e-05,
+      "loss": 0.8663,
+      "step": 183220
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.608238000213262e-05,
+      "loss": 0.9173,
+      "step": 183230
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.608161835270462e-05,
+      "loss": 0.8796,
+      "step": 183240
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.608085670327662e-05,
+      "loss": 0.9408,
+      "step": 183250
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6080095053848615e-05,
+      "loss": 0.8096,
+      "step": 183260
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.607933340442062e-05,
+      "loss": 0.9988,
+      "step": 183270
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.607857175499261e-05,
+      "loss": 1.015,
+      "step": 183280
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6077810105564614e-05,
+      "loss": 0.8651,
+      "step": 183290
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6077048456136605e-05,
+      "loss": 0.997,
+      "step": 183300
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.607628680670861e-05,
+      "loss": 0.9301,
+      "step": 183310
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.607552515728061e-05,
+      "loss": 0.9144,
+      "step": 183320
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6074763507852604e-05,
+      "loss": 0.8524,
+      "step": 183330
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.607400185842461e-05,
+      "loss": 0.88,
+      "step": 183340
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6073240208996606e-05,
+      "loss": 0.8925,
+      "step": 183350
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6072478559568604e-05,
+      "loss": 0.8658,
+      "step": 183360
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.60717169101406e-05,
+      "loss": 0.8444,
+      "step": 183370
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6070955260712605e-05,
+      "loss": 0.9804,
+      "step": 183380
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.6070193611284596e-05,
+      "loss": 1.0065,
+      "step": 183390
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.60694319618566e-05,
+      "loss": 0.7675,
+      "step": 183400
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.60686703124286e-05,
+      "loss": 0.8792,
+      "step": 183410
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.6067908663000595e-05,
+      "loss": 0.9519,
+      "step": 183420
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606714701357259e-05,
+      "loss": 0.8334,
+      "step": 183430
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.60663853641446e-05,
+      "loss": 0.8635,
+      "step": 183440
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606562371471659e-05,
+      "loss": 0.8802,
+      "step": 183450
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606486206528859e-05,
+      "loss": 0.8304,
+      "step": 183460
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606410041586059e-05,
+      "loss": 0.9394,
+      "step": 183470
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606333876643259e-05,
+      "loss": 0.8764,
+      "step": 183480
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.6062577117004584e-05,
+      "loss": 0.9469,
+      "step": 183490
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606181546757659e-05,
+      "loss": 0.8349,
+      "step": 183500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606105381814858e-05,
+      "loss": 0.9035,
+      "step": 183510
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.606029216872058e-05,
+      "loss": 0.9045,
+      "step": 183520
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605953051929258e-05,
+      "loss": 0.8575,
+      "step": 183530
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605876886986458e-05,
+      "loss": 0.8392,
+      "step": 183540
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605800722043658e-05,
+      "loss": 0.9232,
+      "step": 183550
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605724557100858e-05,
+      "loss": 0.7643,
+      "step": 183560
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605648392158058e-05,
+      "loss": 0.9056,
+      "step": 183570
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.6055722272152575e-05,
+      "loss": 0.9042,
+      "step": 183580
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605496062272458e-05,
+      "loss": 0.9956,
+      "step": 183590
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605419897329657e-05,
+      "loss": 0.8407,
+      "step": 183600
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.6053437323868574e-05,
+      "loss": 0.9234,
+      "step": 183610
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.605267567444057e-05,
+      "loss": 0.822,
+      "step": 183620
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.605191402501257e-05,
+      "loss": 0.8983,
+      "step": 183630
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6051152375584566e-05,
+      "loss": 0.8272,
+      "step": 183640
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.605039072615657e-05,
+      "loss": 1.0113,
+      "step": 183650
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604962907672856e-05,
+      "loss": 0.9295,
+      "step": 183660
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6048867427300566e-05,
+      "loss": 0.8596,
+      "step": 183670
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604810577787256e-05,
+      "loss": 0.8408,
+      "step": 183680
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604734412844456e-05,
+      "loss": 0.9972,
+      "step": 183690
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604658247901656e-05,
+      "loss": 0.8192,
+      "step": 183700
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604582082958856e-05,
+      "loss": 0.856,
+      "step": 183710
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604505918016056e-05,
+      "loss": 0.8858,
+      "step": 183720
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604429753073256e-05,
+      "loss": 0.7472,
+      "step": 183730
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6043535881304555e-05,
+      "loss": 0.9853,
+      "step": 183740
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604277423187655e-05,
+      "loss": 0.9617,
+      "step": 183750
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6042012582448556e-05,
+      "loss": 0.838,
+      "step": 183760
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604125093302055e-05,
+      "loss": 0.8962,
+      "step": 183770
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.604048928359255e-05,
+      "loss": 0.9655,
+      "step": 183780
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.603972763416455e-05,
+      "loss": 0.8915,
+      "step": 183790
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6038965984736546e-05,
+      "loss": 0.9769,
+      "step": 183800
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6038204335308544e-05,
+      "loss": 0.8764,
+      "step": 183810
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.603744268588055e-05,
+      "loss": 0.9045,
+      "step": 183820
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.603668103645254e-05,
+      "loss": 0.8529,
+      "step": 183830
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.603591938702454e-05,
+      "loss": 0.8975,
+      "step": 183840
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.603515773759654e-05,
+      "loss": 0.8353,
+      "step": 183850
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.603439608816854e-05,
+      "loss": 0.9477,
+      "step": 183860
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.6033634438740535e-05,
+      "loss": 0.9089,
+      "step": 183870
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.603287278931254e-05,
+      "loss": 0.8947,
+      "step": 183880
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.603211113988453e-05,
+      "loss": 0.8023,
+      "step": 183890
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.6031349490456534e-05,
+      "loss": 0.8896,
+      "step": 183900
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.603058784102853e-05,
+      "loss": 0.9268,
+      "step": 183910
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602982619160053e-05,
+      "loss": 0.9197,
+      "step": 183920
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.6029064542172534e-05,
+      "loss": 0.7999,
+      "step": 183930
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602830289274453e-05,
+      "loss": 0.9031,
+      "step": 183940
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602754124331653e-05,
+      "loss": 0.8558,
+      "step": 183950
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.6026779593888526e-05,
+      "loss": 0.9353,
+      "step": 183960
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602601794446053e-05,
+      "loss": 0.9435,
+      "step": 183970
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602525629503252e-05,
+      "loss": 0.8701,
+      "step": 183980
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.6024494645604525e-05,
+      "loss": 0.8364,
+      "step": 183990
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602373299617652e-05,
+      "loss": 0.911,
+      "step": 184000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602297134674852e-05,
+      "loss": 0.8698,
+      "step": 184010
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602220969732052e-05,
+      "loss": 0.8424,
+      "step": 184020
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602144804789252e-05,
+      "loss": 0.8986,
+      "step": 184030
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.602068639846451e-05,
+      "loss": 0.8994,
+      "step": 184040
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.601992474903652e-05,
+      "loss": 0.8553,
+      "step": 184050
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.6019163099608514e-05,
+      "loss": 0.8663,
+      "step": 184060
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.601840145018051e-05,
+      "loss": 0.9206,
+      "step": 184070
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.601763980075251e-05,
+      "loss": 0.9432,
+      "step": 184080
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6016878151324513e-05,
+      "loss": 0.9806,
+      "step": 184090
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6016116501896504e-05,
+      "loss": 0.8364,
+      "step": 184100
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.601535485246851e-05,
+      "loss": 0.8525,
+      "step": 184110
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6014593203040506e-05,
+      "loss": 0.9923,
+      "step": 184120
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.60138315536125e-05,
+      "loss": 0.9548,
+      "step": 184130
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.601306990418451e-05,
+      "loss": 0.8734,
+      "step": 184140
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6012308254756505e-05,
+      "loss": 0.8476,
+      "step": 184150
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.60115466053285e-05,
+      "loss": 0.8664,
+      "step": 184160
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.60107849559005e-05,
+      "loss": 0.8678,
+      "step": 184170
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6010023306472504e-05,
+      "loss": 0.8735,
+      "step": 184180
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6009261657044495e-05,
+      "loss": 0.8741,
+      "step": 184190
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.60085000076165e-05,
+      "loss": 0.8725,
+      "step": 184200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.60077383581885e-05,
+      "loss": 0.8943,
+      "step": 184210
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6006976708760494e-05,
+      "loss": 0.8696,
+      "step": 184220
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.600621505933249e-05,
+      "loss": 1.0431,
+      "step": 184230
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.600545340990449e-05,
+      "loss": 0.7962,
+      "step": 184240
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6004691760476486e-05,
+      "loss": 0.8579,
+      "step": 184250
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.600393011104849e-05,
+      "loss": 0.9904,
+      "step": 184260
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.600316846162048e-05,
+      "loss": 0.9506,
+      "step": 184270
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6002406812192486e-05,
+      "loss": 0.8753,
+      "step": 184280
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.600164516276448e-05,
+      "loss": 0.914,
+      "step": 184290
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.600088351333648e-05,
+      "loss": 0.9398,
+      "step": 184300
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.600012186390848e-05,
+      "loss": 0.7499,
+      "step": 184310
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599936021448048e-05,
+      "loss": 0.9465,
+      "step": 184320
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599859856505248e-05,
+      "loss": 0.9983,
+      "step": 184330
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599783691562448e-05,
+      "loss": 0.841,
+      "step": 184340
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599707526619648e-05,
+      "loss": 0.8178,
+      "step": 184350
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599631361676847e-05,
+      "loss": 0.8035,
+      "step": 184360
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.5995551967340476e-05,
+      "loss": 0.8532,
+      "step": 184370
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.5994790317912474e-05,
+      "loss": 0.8755,
+      "step": 184380
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599402866848447e-05,
+      "loss": 0.9223,
+      "step": 184390
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599326701905647e-05,
+      "loss": 0.9574,
+      "step": 184400
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599250536962847e-05,
+      "loss": 0.9,
+      "step": 184410
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.5991743720200464e-05,
+      "loss": 0.8916,
+      "step": 184420
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.599098207077247e-05,
+      "loss": 0.8437,
+      "step": 184430
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.5990220421344465e-05,
+      "loss": 0.8929,
+      "step": 184440
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.598945877191646e-05,
+      "loss": 0.842,
+      "step": 184450
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.598869712248846e-05,
+      "loss": 0.9337,
+      "step": 184460
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.5987935473060465e-05,
+      "loss": 0.8269,
+      "step": 184470
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.5987173823632455e-05,
+      "loss": 0.7957,
+      "step": 184480
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.598641217420446e-05,
+      "loss": 0.9543,
+      "step": 184490
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.598565052477646e-05,
+      "loss": 0.8779,
+      "step": 184500
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5984888875348455e-05,
+      "loss": 0.8302,
+      "step": 184510
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.598412722592046e-05,
+      "loss": 0.9018,
+      "step": 184520
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5983365576492456e-05,
+      "loss": 0.9506,
+      "step": 184530
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5982603927064454e-05,
+      "loss": 0.8686,
+      "step": 184540
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.598184227763645e-05,
+      "loss": 0.9705,
+      "step": 184550
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5981080628208455e-05,
+      "loss": 0.9132,
+      "step": 184560
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5980318978780446e-05,
+      "loss": 0.8288,
+      "step": 184570
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597955732935245e-05,
+      "loss": 0.8793,
+      "step": 184580
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597879567992445e-05,
+      "loss": 0.9137,
+      "step": 184590
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5978034030496445e-05,
+      "loss": 0.9521,
+      "step": 184600
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597727238106844e-05,
+      "loss": 1.0063,
+      "step": 184610
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597651073164045e-05,
+      "loss": 0.8754,
+      "step": 184620
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597574908221244e-05,
+      "loss": 0.7938,
+      "step": 184630
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597498743278444e-05,
+      "loss": 0.8737,
+      "step": 184640
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597422578335644e-05,
+      "loss": 0.833,
+      "step": 184650
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597346413392844e-05,
+      "loss": 0.866,
+      "step": 184660
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5972702484500434e-05,
+      "loss": 0.8597,
+      "step": 184670
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597194083507243e-05,
+      "loss": 0.9219,
+      "step": 184680
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.597117918564443e-05,
+      "loss": 0.8157,
+      "step": 184690
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.5970417536216433e-05,
+      "loss": 0.8673,
+      "step": 184700
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.596965588678843e-05,
+      "loss": 0.8753,
+      "step": 184710
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.596889423736043e-05,
+      "loss": 0.8442,
+      "step": 184720
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596813258793243e-05,
+      "loss": 0.8361,
+      "step": 184730
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596737093850442e-05,
+      "loss": 0.8869,
+      "step": 184740
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596660928907643e-05,
+      "loss": 0.9167,
+      "step": 184750
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5965847639648425e-05,
+      "loss": 0.8758,
+      "step": 184760
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596508599022042e-05,
+      "loss": 0.9532,
+      "step": 184770
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596432434079242e-05,
+      "loss": 0.991,
+      "step": 184780
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5963562691364424e-05,
+      "loss": 0.9201,
+      "step": 184790
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5962801041936415e-05,
+      "loss": 0.9062,
+      "step": 184800
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596203939250842e-05,
+      "loss": 0.9683,
+      "step": 184810
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.596127774308042e-05,
+      "loss": 0.9234,
+      "step": 184820
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5960516093652414e-05,
+      "loss": 0.8637,
+      "step": 184830
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.595975444422441e-05,
+      "loss": 0.9124,
+      "step": 184840
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5958992794796416e-05,
+      "loss": 0.8761,
+      "step": 184850
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5958231145368406e-05,
+      "loss": 0.9568,
+      "step": 184860
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.595746949594041e-05,
+      "loss": 0.9638,
+      "step": 184870
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.595670784651241e-05,
+      "loss": 0.8302,
+      "step": 184880
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5955946197084406e-05,
+      "loss": 0.8199,
+      "step": 184890
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.59551845476564e-05,
+      "loss": 0.8542,
+      "step": 184900
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.595442289822841e-05,
+      "loss": 0.8704,
+      "step": 184910
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5953661248800405e-05,
+      "loss": 0.8145,
+      "step": 184920
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.59528995993724e-05,
+      "loss": 0.9053,
+      "step": 184930
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.5952137949944407e-05,
+      "loss": 0.9937,
+      "step": 184940
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.59513763005164e-05,
+      "loss": 0.9269,
+      "step": 184950
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.59506146510884e-05,
+      "loss": 0.8898,
+      "step": 184960
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.59498530016604e-05,
+      "loss": 0.9017,
+      "step": 184970
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.5949091352232396e-05,
+      "loss": 0.8086,
+      "step": 184980
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.5948329702804394e-05,
+      "loss": 0.8299,
+      "step": 184990
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.59475680533764e-05,
+      "loss": 0.9379,
+      "step": 185000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594680640394839e-05,
+      "loss": 0.8786,
+      "step": 185010
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594604475452039e-05,
+      "loss": 0.9575,
+      "step": 185020
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594528310509239e-05,
+      "loss": 0.8536,
+      "step": 185030
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594452145566439e-05,
+      "loss": 0.8721,
+      "step": 185040
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.5943759806236385e-05,
+      "loss": 0.8914,
+      "step": 185050
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594299815680839e-05,
+      "loss": 0.8763,
+      "step": 185060
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594223650738038e-05,
+      "loss": 0.9117,
+      "step": 185070
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.5941474857952385e-05,
+      "loss": 0.8762,
+      "step": 185080
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.594071320852438e-05,
+      "loss": 0.8812,
+      "step": 185090
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.593995155909638e-05,
+      "loss": 0.8227,
+      "step": 185100
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.593918990966838e-05,
+      "loss": 0.9269,
+      "step": 185110
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.593842826024038e-05,
+      "loss": 0.9461,
+      "step": 185120
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.593766661081238e-05,
+      "loss": 0.8659,
+      "step": 185130
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.5936904961384376e-05,
+      "loss": 0.8282,
+      "step": 185140
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.5936143311956374e-05,
+      "loss": 0.8548,
+      "step": 185150
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.593538166252837e-05,
+      "loss": 1.0474,
+      "step": 185160
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5934620013100375e-05,
+      "loss": 0.8068,
+      "step": 185170
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5933858363672366e-05,
+      "loss": 0.8733,
+      "step": 185180
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.593309671424437e-05,
+      "loss": 0.9649,
+      "step": 185190
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.593233506481637e-05,
+      "loss": 0.8268,
+      "step": 185200
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5931573415388365e-05,
+      "loss": 0.9013,
+      "step": 185210
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.593081176596036e-05,
+      "loss": 0.8748,
+      "step": 185220
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.593005011653237e-05,
+      "loss": 0.8574,
+      "step": 185230
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592928846710436e-05,
+      "loss": 0.9424,
+      "step": 185240
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592852681767636e-05,
+      "loss": 0.8989,
+      "step": 185250
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592776516824836e-05,
+      "loss": 0.9693,
+      "step": 185260
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592700351882036e-05,
+      "loss": 0.8306,
+      "step": 185270
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5926241869392354e-05,
+      "loss": 0.8377,
+      "step": 185280
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592548021996436e-05,
+      "loss": 0.9137,
+      "step": 185290
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5924718570536356e-05,
+      "loss": 0.8841,
+      "step": 185300
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5923956921108353e-05,
+      "loss": 0.9519,
+      "step": 185310
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592319527168036e-05,
+      "loss": 0.9051,
+      "step": 185320
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592243362225235e-05,
+      "loss": 0.8692,
+      "step": 185330
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592167197282435e-05,
+      "loss": 0.9396,
+      "step": 185340
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592091032339635e-05,
+      "loss": 0.8998,
+      "step": 185350
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.592014867396835e-05,
+      "loss": 1.0015,
+      "step": 185360
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.5919387024540345e-05,
+      "loss": 0.8833,
+      "step": 185370
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.591862537511235e-05,
+      "loss": 0.8674,
+      "step": 185380
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591786372568434e-05,
+      "loss": 0.973,
+      "step": 185390
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.5917102076256344e-05,
+      "loss": 0.8999,
+      "step": 185400
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591634042682834e-05,
+      "loss": 0.772,
+      "step": 185410
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591557877740034e-05,
+      "loss": 0.8165,
+      "step": 185420
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591481712797234e-05,
+      "loss": 0.8753,
+      "step": 185430
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591405547854434e-05,
+      "loss": 0.8955,
+      "step": 185440
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591329382911633e-05,
+      "loss": 0.843,
+      "step": 185450
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.5912532179688336e-05,
+      "loss": 0.898,
+      "step": 185460
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591177053026033e-05,
+      "loss": 0.8559,
+      "step": 185470
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591100888083233e-05,
+      "loss": 0.8879,
+      "step": 185480
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.591024723140433e-05,
+      "loss": 0.7924,
+      "step": 185490
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590948558197633e-05,
+      "loss": 0.9527,
+      "step": 185500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590872393254833e-05,
+      "loss": 0.851,
+      "step": 185510
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590796228312033e-05,
+      "loss": 0.8578,
+      "step": 185520
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590720063369233e-05,
+      "loss": 0.9362,
+      "step": 185530
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590643898426432e-05,
+      "loss": 0.9094,
+      "step": 185540
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.5905677334836327e-05,
+      "loss": 0.8445,
+      "step": 185550
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.5904915685408324e-05,
+      "loss": 0.9196,
+      "step": 185560
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590415403598032e-05,
+      "loss": 0.8638,
+      "step": 185570
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590339238655232e-05,
+      "loss": 0.8742,
+      "step": 185580
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.590263073712432e-05,
+      "loss": 0.8644,
+      "step": 185590
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.5901869087696314e-05,
+      "loss": 0.8003,
+      "step": 185600
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.590110743826832e-05,
+      "loss": 0.8355,
+      "step": 185610
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.590034578884031e-05,
+      "loss": 0.8814,
+      "step": 185620
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.589958413941231e-05,
+      "loss": 0.9674,
+      "step": 185630
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.589882248998431e-05,
+      "loss": 0.8511,
+      "step": 185640
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.589806084055631e-05,
+      "loss": 0.901,
+      "step": 185650
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.5897299191128305e-05,
+      "loss": 0.9358,
+      "step": 185660
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.589653754170031e-05,
+      "loss": 0.8946,
+      "step": 185670
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.589577589227231e-05,
+      "loss": 0.8949,
+      "step": 185680
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.5895014242844305e-05,
+      "loss": 0.9096,
+      "step": 185690
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.58942525934163e-05,
+      "loss": 0.8891,
+      "step": 185700
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.58934909439883e-05,
+      "loss": 0.8916,
+      "step": 185710
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.5892729294560304e-05,
+      "loss": 0.8302,
+      "step": 185720
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.58919676451323e-05,
+      "loss": 0.825,
+      "step": 185730
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.58912059957043e-05,
+      "loss": 0.9758,
+      "step": 185740
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.5890444346276296e-05,
+      "loss": 0.8581,
+      "step": 185750
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.58896826968483e-05,
+      "loss": 0.8898,
+      "step": 185760
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.588892104742029e-05,
+      "loss": 0.9727,
+      "step": 185770
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.5888159397992295e-05,
+      "loss": 0.8096,
+      "step": 185780
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.588739774856429e-05,
+      "loss": 0.7575,
+      "step": 185790
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.588663609913629e-05,
+      "loss": 0.8843,
+      "step": 185800
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.588587444970829e-05,
+      "loss": 0.8434,
+      "step": 185810
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.588511280028029e-05,
+      "loss": 0.9881,
+      "step": 185820
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.588435115085228e-05,
+      "loss": 0.7151,
+      "step": 185830
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.588358950142429e-05,
+      "loss": 0.8423,
+      "step": 185840
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5882827851996284e-05,
+      "loss": 0.8841,
+      "step": 185850
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.588206620256828e-05,
+      "loss": 0.9485,
+      "step": 185860
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.588130455314028e-05,
+      "loss": 0.8821,
+      "step": 185870
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5880542903712284e-05,
+      "loss": 1.065,
+      "step": 185880
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587978125428428e-05,
+      "loss": 0.8806,
+      "step": 185890
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587901960485628e-05,
+      "loss": 0.9821,
+      "step": 185900
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5878257955428276e-05,
+      "loss": 0.8299,
+      "step": 185910
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5877496306000273e-05,
+      "loss": 0.8307,
+      "step": 185920
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587673465657228e-05,
+      "loss": 0.9476,
+      "step": 185930
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5875973007144275e-05,
+      "loss": 0.9827,
+      "step": 185940
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587521135771627e-05,
+      "loss": 1.0247,
+      "step": 185950
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587444970828827e-05,
+      "loss": 0.8857,
+      "step": 185960
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5873688058860274e-05,
+      "loss": 0.9455,
+      "step": 185970
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5872926409432265e-05,
+      "loss": 0.8772,
+      "step": 185980
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587216476000427e-05,
+      "loss": 0.8199,
+      "step": 185990
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.587140311057627e-05,
+      "loss": 0.8464,
+      "step": 186000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5870641461148264e-05,
+      "loss": 0.9084,
+      "step": 186010
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.586987981172026e-05,
+      "loss": 1.0017,
+      "step": 186020
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.5869118162292266e-05,
+      "loss": 0.8499,
+      "step": 186030
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586835651286426e-05,
+      "loss": 0.8743,
+      "step": 186040
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586759486343626e-05,
+      "loss": 0.8905,
+      "step": 186050
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586683321400826e-05,
+      "loss": 0.8984,
+      "step": 186060
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.5866071564580256e-05,
+      "loss": 0.8497,
+      "step": 186070
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586530991515225e-05,
+      "loss": 0.8293,
+      "step": 186080
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586454826572425e-05,
+      "loss": 0.9892,
+      "step": 186090
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.5863786616296255e-05,
+      "loss": 0.8805,
+      "step": 186100
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586302496686825e-05,
+      "loss": 0.9692,
+      "step": 186110
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586226331744025e-05,
+      "loss": 0.8339,
+      "step": 186120
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586150166801225e-05,
+      "loss": 0.8629,
+      "step": 186130
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.586074001858425e-05,
+      "loss": 0.967,
+      "step": 186140
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585997836915624e-05,
+      "loss": 0.9561,
+      "step": 186150
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.5859216719728247e-05,
+      "loss": 0.9509,
+      "step": 186160
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.5858455070300244e-05,
+      "loss": 0.9512,
+      "step": 186170
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585769342087224e-05,
+      "loss": 0.9156,
+      "step": 186180
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585693177144424e-05,
+      "loss": 0.8246,
+      "step": 186190
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585617012201624e-05,
+      "loss": 0.9237,
+      "step": 186200
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.5855408472588234e-05,
+      "loss": 0.8969,
+      "step": 186210
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585464682316024e-05,
+      "loss": 0.9089,
+      "step": 186220
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.5853885173732236e-05,
+      "loss": 0.901,
+      "step": 186230
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585312352430423e-05,
+      "loss": 0.8655,
+      "step": 186240
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.585236187487623e-05,
+      "loss": 0.9283,
+      "step": 186250
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5851600225448235e-05,
+      "loss": 0.7763,
+      "step": 186260
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.585083857602023e-05,
+      "loss": 1.0,
+      "step": 186270
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.585007692659223e-05,
+      "loss": 0.8958,
+      "step": 186280
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584931527716423e-05,
+      "loss": 0.7624,
+      "step": 186290
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5848553627736225e-05,
+      "loss": 0.863,
+      "step": 186300
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584779197830823e-05,
+      "loss": 0.8959,
+      "step": 186310
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5847030328880226e-05,
+      "loss": 0.8951,
+      "step": 186320
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5846268679452224e-05,
+      "loss": 0.9595,
+      "step": 186330
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584550703002422e-05,
+      "loss": 0.8518,
+      "step": 186340
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5844745380596226e-05,
+      "loss": 0.947,
+      "step": 186350
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5843983731168216e-05,
+      "loss": 0.8877,
+      "step": 186360
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584322208174022e-05,
+      "loss": 0.926,
+      "step": 186370
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584246043231222e-05,
+      "loss": 0.8789,
+      "step": 186380
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5841698782884215e-05,
+      "loss": 0.8913,
+      "step": 186390
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584093713345621e-05,
+      "loss": 0.9233,
+      "step": 186400
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.584017548402822e-05,
+      "loss": 0.8808,
+      "step": 186410
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.583941383460021e-05,
+      "loss": 0.8716,
+      "step": 186420
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.583865218517221e-05,
+      "loss": 0.8713,
+      "step": 186430
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.583789053574421e-05,
+      "loss": 0.8085,
+      "step": 186440
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.583712888631621e-05,
+      "loss": 0.9179,
+      "step": 186450
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.5836367236888204e-05,
+      "loss": 0.8275,
+      "step": 186460
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.583560558746021e-05,
+      "loss": 0.9817,
+      "step": 186470
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5834843938032206e-05,
+      "loss": 0.8574,
+      "step": 186480
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5834082288604204e-05,
+      "loss": 0.8799,
+      "step": 186490
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.58333206391762e-05,
+      "loss": 1.015,
+      "step": 186500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.58325589897482e-05,
+      "loss": 0.9636,
+      "step": 186510
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.58317973403202e-05,
+      "loss": 0.8695,
+      "step": 186520
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5831035690892194e-05,
+      "loss": 0.8099,
+      "step": 186530
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.58302740414642e-05,
+      "loss": 0.9836,
+      "step": 186540
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5829512392036195e-05,
+      "loss": 0.937,
+      "step": 186550
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582875074260819e-05,
+      "loss": 0.8742,
+      "step": 186560
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582798909318019e-05,
+      "loss": 0.8547,
+      "step": 186570
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5827227443752194e-05,
+      "loss": 0.9167,
+      "step": 186580
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5826465794324185e-05,
+      "loss": 0.9321,
+      "step": 186590
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582570414489619e-05,
+      "loss": 0.8783,
+      "step": 186600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582494249546819e-05,
+      "loss": 0.7986,
+      "step": 186610
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5824180846040184e-05,
+      "loss": 0.8918,
+      "step": 186620
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582341919661218e-05,
+      "loss": 0.9726,
+      "step": 186630
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5822657547184186e-05,
+      "loss": 1.0072,
+      "step": 186640
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5821895897756183e-05,
+      "loss": 1.076,
+      "step": 186650
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582113424832818e-05,
+      "loss": 0.9352,
+      "step": 186660
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.582037259890018e-05,
+      "loss": 0.8249,
+      "step": 186670
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.5819610949472176e-05,
+      "loss": 0.8535,
+      "step": 186680
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.581884930004418e-05,
+      "loss": 0.9184,
+      "step": 186690
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581808765061618e-05,
+      "loss": 0.8795,
+      "step": 186700
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.5817326001188175e-05,
+      "loss": 0.9535,
+      "step": 186710
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581656435176017e-05,
+      "loss": 0.8688,
+      "step": 186720
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581580270233218e-05,
+      "loss": 0.9003,
+      "step": 186730
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581504105290417e-05,
+      "loss": 0.8323,
+      "step": 186740
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581427940347617e-05,
+      "loss": 0.912,
+      "step": 186750
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581351775404817e-05,
+      "loss": 0.8586,
+      "step": 186760
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.5812756104620167e-05,
+      "loss": 0.9896,
+      "step": 186770
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.5811994455192164e-05,
+      "loss": 0.8522,
+      "step": 186780
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581123280576417e-05,
+      "loss": 0.856,
+      "step": 186790
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.581047115633616e-05,
+      "loss": 0.9404,
+      "step": 186800
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580970950690816e-05,
+      "loss": 0.9974,
+      "step": 186810
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580894785748016e-05,
+      "loss": 0.8611,
+      "step": 186820
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580818620805216e-05,
+      "loss": 0.8767,
+      "step": 186830
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.5807424558624156e-05,
+      "loss": 0.8779,
+      "step": 186840
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580666290919616e-05,
+      "loss": 0.8172,
+      "step": 186850
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580590125976816e-05,
+      "loss": 0.9117,
+      "step": 186860
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.5805139610340155e-05,
+      "loss": 0.9185,
+      "step": 186870
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580437796091215e-05,
+      "loss": 0.7839,
+      "step": 186880
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580361631148415e-05,
+      "loss": 0.8171,
+      "step": 186890
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.5802854662056154e-05,
+      "loss": 0.863,
+      "step": 186900
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.580209301262815e-05,
+      "loss": 0.9431,
+      "step": 186910
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.580133136320015e-05,
+      "loss": 0.8695,
+      "step": 186920
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.5800569713772146e-05,
+      "loss": 0.9,
+      "step": 186930
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579980806434415e-05,
+      "loss": 0.7805,
+      "step": 186940
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579904641491614e-05,
+      "loss": 0.9179,
+      "step": 186950
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.5798284765488146e-05,
+      "loss": 0.9735,
+      "step": 186960
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579752311606014e-05,
+      "loss": 0.8866,
+      "step": 186970
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579676146663214e-05,
+      "loss": 0.9543,
+      "step": 186980
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579599981720414e-05,
+      "loss": 0.9227,
+      "step": 186990
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.5795238167776135e-05,
+      "loss": 0.8827,
+      "step": 187000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579447651834813e-05,
+      "loss": 0.8992,
+      "step": 187010
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579371486892014e-05,
+      "loss": 0.8602,
+      "step": 187020
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.5792953219492135e-05,
+      "loss": 0.8344,
+      "step": 187030
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579219157006413e-05,
+      "loss": 0.9653,
+      "step": 187040
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579142992063613e-05,
+      "loss": 0.8847,
+      "step": 187050
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.579066827120813e-05,
+      "loss": 1.0763,
+      "step": 187060
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.578990662178013e-05,
+      "loss": 0.9375,
+      "step": 187070
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.578914497235213e-05,
+      "loss": 0.9006,
+      "step": 187080
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.5788383322924126e-05,
+      "loss": 0.9168,
+      "step": 187090
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.5787621673496124e-05,
+      "loss": 0.8698,
+      "step": 187100
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.578686002406813e-05,
+      "loss": 0.8794,
+      "step": 187110
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.578609837464012e-05,
+      "loss": 0.7607,
+      "step": 187120
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.578533672521212e-05,
+      "loss": 0.8741,
+      "step": 187130
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.578457507578412e-05,
+      "loss": 0.8853,
+      "step": 187140
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.578381342635612e-05,
+      "loss": 0.93,
+      "step": 187150
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.5783051776928115e-05,
+      "loss": 0.9022,
+      "step": 187160
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.578229012750012e-05,
+      "loss": 0.9177,
+      "step": 187170
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.578152847807211e-05,
+      "loss": 0.8469,
+      "step": 187180
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.5780766828644114e-05,
+      "loss": 0.8131,
+      "step": 187190
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.578000517921611e-05,
+      "loss": 0.8557,
+      "step": 187200
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.577924352978811e-05,
+      "loss": 0.8247,
+      "step": 187210
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.577848188036011e-05,
+      "loss": 0.8139,
+      "step": 187220
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.577772023093211e-05,
+      "loss": 0.863,
+      "step": 187230
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.577695858150411e-05,
+      "loss": 0.8846,
+      "step": 187240
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.5776196932076106e-05,
+      "loss": 0.8253,
+      "step": 187250
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.5775435282648103e-05,
+      "loss": 0.9276,
+      "step": 187260
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.57746736332201e-05,
+      "loss": 0.956,
+      "step": 187270
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.5773911983792105e-05,
+      "loss": 0.9399,
+      "step": 187280
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.57731503343641e-05,
+      "loss": 0.8178,
+      "step": 187290
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.57723886849361e-05,
+      "loss": 0.8866,
+      "step": 187300
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.57716270355081e-05,
+      "loss": 0.9838,
+      "step": 187310
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.57708653860801e-05,
+      "loss": 0.8987,
+      "step": 187320
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.577010373665209e-05,
+      "loss": 0.8714,
+      "step": 187330
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.57693420872241e-05,
+      "loss": 0.8574,
+      "step": 187340
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.5768580437796094e-05,
+      "loss": 0.9387,
+      "step": 187350
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576781878836809e-05,
+      "loss": 0.974,
+      "step": 187360
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576705713894009e-05,
+      "loss": 1.0367,
+      "step": 187370
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576629548951209e-05,
+      "loss": 0.8653,
+      "step": 187380
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.5765533840084084e-05,
+      "loss": 0.9084,
+      "step": 187390
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576477219065609e-05,
+      "loss": 0.883,
+      "step": 187400
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.5764010541228086e-05,
+      "loss": 0.8721,
+      "step": 187410
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576324889180008e-05,
+      "loss": 0.9483,
+      "step": 187420
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576248724237208e-05,
+      "loss": 0.8179,
+      "step": 187430
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576172559294408e-05,
+      "loss": 0.8871,
+      "step": 187440
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576096394351608e-05,
+      "loss": 0.8437,
+      "step": 187450
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.576020229408808e-05,
+      "loss": 0.9524,
+      "step": 187460
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575944064466008e-05,
+      "loss": 0.8744,
+      "step": 187470
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.5758678995232075e-05,
+      "loss": 0.9492,
+      "step": 187480
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575791734580408e-05,
+      "loss": 0.8864,
+      "step": 187490
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575715569637607e-05,
+      "loss": 0.9019,
+      "step": 187500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.5756394046948074e-05,
+      "loss": 0.8733,
+      "step": 187510
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575563239752007e-05,
+      "loss": 0.9185,
+      "step": 187520
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575487074809207e-05,
+      "loss": 0.9032,
+      "step": 187530
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.5754109098664066e-05,
+      "loss": 0.9895,
+      "step": 187540
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575334744923607e-05,
+      "loss": 0.8636,
+      "step": 187550
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.575258579980806e-05,
+      "loss": 0.9254,
+      "step": 187560
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.5751824150380066e-05,
+      "loss": 0.891,
+      "step": 187570
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.575106250095206e-05,
+      "loss": 0.8421,
+      "step": 187580
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.575030085152406e-05,
+      "loss": 0.852,
+      "step": 187590
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574953920209606e-05,
+      "loss": 0.8942,
+      "step": 187600
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574877755266806e-05,
+      "loss": 0.83,
+      "step": 187610
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574801590324006e-05,
+      "loss": 0.873,
+      "step": 187620
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574725425381206e-05,
+      "loss": 0.8835,
+      "step": 187630
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5746492604384055e-05,
+      "loss": 0.8793,
+      "step": 187640
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574573095495605e-05,
+      "loss": 0.9153,
+      "step": 187650
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5744969305528056e-05,
+      "loss": 0.8162,
+      "step": 187660
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5744207656100054e-05,
+      "loss": 0.9079,
+      "step": 187670
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574344600667205e-05,
+      "loss": 0.88,
+      "step": 187680
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574268435724405e-05,
+      "loss": 0.9623,
+      "step": 187690
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574192270781605e-05,
+      "loss": 0.9335,
+      "step": 187700
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5741161058388044e-05,
+      "loss": 0.9041,
+      "step": 187710
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.574039940896005e-05,
+      "loss": 0.906,
+      "step": 187720
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5739637759532045e-05,
+      "loss": 0.8833,
+      "step": 187730
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.573887611010404e-05,
+      "loss": 1.0084,
+      "step": 187740
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.573811446067604e-05,
+      "loss": 0.9225,
+      "step": 187750
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5737352811248045e-05,
+      "loss": 0.8934,
+      "step": 187760
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.5736591161820035e-05,
+      "loss": 0.9712,
+      "step": 187770
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.573582951239204e-05,
+      "loss": 0.8748,
+      "step": 187780
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.573506786296404e-05,
+      "loss": 0.9697,
+      "step": 187790
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5734306213536034e-05,
+      "loss": 0.8576,
+      "step": 187800
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.573354456410803e-05,
+      "loss": 0.9326,
+      "step": 187810
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5732782914680036e-05,
+      "loss": 0.8836,
+      "step": 187820
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5732021265252034e-05,
+      "loss": 0.9804,
+      "step": 187830
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.573125961582403e-05,
+      "loss": 0.7828,
+      "step": 187840
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.573049796639603e-05,
+      "loss": 0.8049,
+      "step": 187850
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5729736316968026e-05,
+      "loss": 0.8445,
+      "step": 187860
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572897466754003e-05,
+      "loss": 0.9247,
+      "step": 187870
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572821301811203e-05,
+      "loss": 0.944,
+      "step": 187880
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5727451368684025e-05,
+      "loss": 0.847,
+      "step": 187890
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572668971925602e-05,
+      "loss": 0.8684,
+      "step": 187900
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572592806982802e-05,
+      "loss": 0.8113,
+      "step": 187910
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572516642040002e-05,
+      "loss": 0.9376,
+      "step": 187920
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572440477097202e-05,
+      "loss": 0.9994,
+      "step": 187930
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572364312154401e-05,
+      "loss": 0.8282,
+      "step": 187940
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572288147211602e-05,
+      "loss": 0.8491,
+      "step": 187950
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5722119822688014e-05,
+      "loss": 0.9302,
+      "step": 187960
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572135817326001e-05,
+      "loss": 0.8727,
+      "step": 187970
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.572059652383201e-05,
+      "loss": 0.845,
+      "step": 187980
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5719834874404013e-05,
+      "loss": 0.8978,
+      "step": 187990
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.5719073224976004e-05,
+      "loss": 0.8491,
+      "step": 188000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.571831157554801e-05,
+      "loss": 0.8569,
+      "step": 188010
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5717549926120006e-05,
+      "loss": 0.9063,
+      "step": 188020
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5716788276692e-05,
+      "loss": 0.9087,
+      "step": 188030
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.571602662726401e-05,
+      "loss": 0.8621,
+      "step": 188040
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5715264977836005e-05,
+      "loss": 0.8729,
+      "step": 188050
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5714503328408e-05,
+      "loss": 0.862,
+      "step": 188060
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.571374167898e-05,
+      "loss": 0.9474,
+      "step": 188070
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5712980029552004e-05,
+      "loss": 0.9056,
+      "step": 188080
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5712218380123995e-05,
+      "loss": 0.879,
+      "step": 188090
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5711456730696e-05,
+      "loss": 0.9033,
+      "step": 188100
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5710695081267997e-05,
+      "loss": 0.8374,
+      "step": 188110
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5709933431839994e-05,
+      "loss": 0.8569,
+      "step": 188120
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570917178241199e-05,
+      "loss": 0.8417,
+      "step": 188130
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5708410132983996e-05,
+      "loss": 0.8441,
+      "step": 188140
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5707648483555986e-05,
+      "loss": 0.9459,
+      "step": 188150
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570688683412799e-05,
+      "loss": 0.9299,
+      "step": 188160
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570612518469999e-05,
+      "loss": 0.976,
+      "step": 188170
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.5705363535271986e-05,
+      "loss": 0.8954,
+      "step": 188180
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570460188584398e-05,
+      "loss": 0.8904,
+      "step": 188190
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570384023641599e-05,
+      "loss": 0.8353,
+      "step": 188200
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570307858698798e-05,
+      "loss": 0.9006,
+      "step": 188210
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.570231693755998e-05,
+      "loss": 0.934,
+      "step": 188220
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.570155528813198e-05,
+      "loss": 0.7737,
+      "step": 188230
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.570079363870398e-05,
+      "loss": 0.8158,
+      "step": 188240
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.570003198927598e-05,
+      "loss": 0.8332,
+      "step": 188250
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569927033984798e-05,
+      "loss": 0.9231,
+      "step": 188260
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5698508690419976e-05,
+      "loss": 0.8407,
+      "step": 188270
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5697747040991974e-05,
+      "loss": 0.881,
+      "step": 188280
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569698539156398e-05,
+      "loss": 0.9207,
+      "step": 188290
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569622374213597e-05,
+      "loss": 0.9104,
+      "step": 188300
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569546209270797e-05,
+      "loss": 0.9053,
+      "step": 188310
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569470044327997e-05,
+      "loss": 0.9501,
+      "step": 188320
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569393879385197e-05,
+      "loss": 0.8818,
+      "step": 188330
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5693177144423965e-05,
+      "loss": 0.7928,
+      "step": 188340
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569241549499597e-05,
+      "loss": 1.0265,
+      "step": 188350
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.569165384556796e-05,
+      "loss": 0.8924,
+      "step": 188360
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5690892196139965e-05,
+      "loss": 0.9658,
+      "step": 188370
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5690130546711955e-05,
+      "loss": 1.0344,
+      "step": 188380
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.568936889728396e-05,
+      "loss": 0.7598,
+      "step": 188390
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.568860724785596e-05,
+      "loss": 0.8582,
+      "step": 188400
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5687845598427954e-05,
+      "loss": 0.9384,
+      "step": 188410
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.568708394899996e-05,
+      "loss": 0.8875,
+      "step": 188420
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5686322299571956e-05,
+      "loss": 0.9313,
+      "step": 188430
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.5685560650143954e-05,
+      "loss": 1.0268,
+      "step": 188440
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.568479900071595e-05,
+      "loss": 0.7936,
+      "step": 188450
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.5684037351287955e-05,
+      "loss": 0.9282,
+      "step": 188460
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.5683275701859946e-05,
+      "loss": 0.8705,
+      "step": 188470
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.568251405243195e-05,
+      "loss": 0.9793,
+      "step": 188480
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.568175240300395e-05,
+      "loss": 0.8629,
+      "step": 188490
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.5680990753575945e-05,
+      "loss": 0.9326,
+      "step": 188500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.568022910414794e-05,
+      "loss": 0.8735,
+      "step": 188510
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567946745471995e-05,
+      "loss": 0.9433,
+      "step": 188520
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567870580529194e-05,
+      "loss": 0.9249,
+      "step": 188530
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567794415586394e-05,
+      "loss": 0.8789,
+      "step": 188540
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567718250643594e-05,
+      "loss": 0.8481,
+      "step": 188550
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567642085700794e-05,
+      "loss": 0.9903,
+      "step": 188560
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.5675659207579934e-05,
+      "loss": 0.807,
+      "step": 188570
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567489755815194e-05,
+      "loss": 0.9133,
+      "step": 188580
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567413590872393e-05,
+      "loss": 0.8638,
+      "step": 188590
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.5673374259295933e-05,
+      "loss": 0.8801,
+      "step": 188600
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567261260986793e-05,
+      "loss": 0.8402,
+      "step": 188610
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567185096043993e-05,
+      "loss": 0.8246,
+      "step": 188620
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567108931101193e-05,
+      "loss": 0.8198,
+      "step": 188630
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.567032766158393e-05,
+      "loss": 0.9022,
+      "step": 188640
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.566956601215593e-05,
+      "loss": 0.8572,
+      "step": 188650
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.5668804362727925e-05,
+      "loss": 0.9382,
+      "step": 188660
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566804271329993e-05,
+      "loss": 0.9089,
+      "step": 188670
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566728106387192e-05,
+      "loss": 0.9174,
+      "step": 188680
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.5666519414443924e-05,
+      "loss": 0.8832,
+      "step": 188690
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566575776501592e-05,
+      "loss": 0.8388,
+      "step": 188700
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566499611558792e-05,
+      "loss": 0.9713,
+      "step": 188710
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.5664234466159917e-05,
+      "loss": 0.8687,
+      "step": 188720
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566347281673192e-05,
+      "loss": 0.9209,
+      "step": 188730
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566271116730391e-05,
+      "loss": 0.9539,
+      "step": 188740
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.5661949517875916e-05,
+      "loss": 0.8691,
+      "step": 188750
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566118786844791e-05,
+      "loss": 0.9087,
+      "step": 188760
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.566042621901991e-05,
+      "loss": 0.8732,
+      "step": 188770
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.565966456959191e-05,
+      "loss": 0.8342,
+      "step": 188780
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.565890292016391e-05,
+      "loss": 0.8995,
+      "step": 188790
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.56581412707359e-05,
+      "loss": 0.9761,
+      "step": 188800
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.565737962130791e-05,
+      "loss": 0.884,
+      "step": 188810
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.5656617971879905e-05,
+      "loss": 0.8774,
+      "step": 188820
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.56558563224519e-05,
+      "loss": 0.9962,
+      "step": 188830
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.5655094673023906e-05,
+      "loss": 1.03,
+      "step": 188840
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.56543330235959e-05,
+      "loss": 0.9039,
+      "step": 188850
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.56535713741679e-05,
+      "loss": 0.8009,
+      "step": 188860
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.56528097247399e-05,
+      "loss": 0.8664,
+      "step": 188870
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.5652048075311896e-05,
+      "loss": 0.9571,
+      "step": 188880
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.5651286425883894e-05,
+      "loss": 0.922,
+      "step": 188890
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.56505247764559e-05,
+      "loss": 0.8271,
+      "step": 188900
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564976312702789e-05,
+      "loss": 0.9063,
+      "step": 188910
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564900147759989e-05,
+      "loss": 0.8828,
+      "step": 188920
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564823982817189e-05,
+      "loss": 0.9298,
+      "step": 188930
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564747817874389e-05,
+      "loss": 0.8213,
+      "step": 188940
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.5646716529315885e-05,
+      "loss": 0.7093,
+      "step": 188950
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564595487988789e-05,
+      "loss": 0.9035,
+      "step": 188960
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564519323045988e-05,
+      "loss": 0.9059,
+      "step": 188970
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.5644431581031885e-05,
+      "loss": 0.8696,
+      "step": 188980
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564366993160388e-05,
+      "loss": 0.9675,
+      "step": 188990
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564290828217588e-05,
+      "loss": 0.8523,
+      "step": 189000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564214663274788e-05,
+      "loss": 0.8686,
+      "step": 189010
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564138498331988e-05,
+      "loss": 0.8507,
+      "step": 189020
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.564062333389188e-05,
+      "loss": 0.9183,
+      "step": 189030
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.5639861684463876e-05,
+      "loss": 0.8544,
+      "step": 189040
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.563910003503588e-05,
+      "loss": 0.8298,
+      "step": 189050
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.563833838560787e-05,
+      "loss": 0.783,
+      "step": 189060
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.5637576736179875e-05,
+      "loss": 0.893,
+      "step": 189070
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.563681508675187e-05,
+      "loss": 0.8396,
+      "step": 189080
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.563605343732387e-05,
+      "loss": 0.8124,
+      "step": 189090
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.563529178789587e-05,
+      "loss": 0.9699,
+      "step": 189100
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.563453013846787e-05,
+      "loss": 0.9216,
+      "step": 189110
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.563376848903986e-05,
+      "loss": 0.907,
+      "step": 189120
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.563300683961187e-05,
+      "loss": 0.8348,
+      "step": 189130
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5632245190183864e-05,
+      "loss": 0.7893,
+      "step": 189140
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.563148354075586e-05,
+      "loss": 0.8775,
+      "step": 189150
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.563072189132786e-05,
+      "loss": 0.8937,
+      "step": 189160
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5629960241899864e-05,
+      "loss": 0.9662,
+      "step": 189170
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5629198592471854e-05,
+      "loss": 0.7818,
+      "step": 189180
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.562843694304386e-05,
+      "loss": 0.9488,
+      "step": 189190
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5627675293615856e-05,
+      "loss": 0.8671,
+      "step": 189200
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5626913644187853e-05,
+      "loss": 0.9032,
+      "step": 189210
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.562615199475986e-05,
+      "loss": 0.8966,
+      "step": 189220
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5625390345331855e-05,
+      "loss": 0.9821,
+      "step": 189230
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.562462869590385e-05,
+      "loss": 0.8728,
+      "step": 189240
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.562386704647585e-05,
+      "loss": 0.8955,
+      "step": 189250
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5623105397047854e-05,
+      "loss": 0.8737,
+      "step": 189260
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5622343747619845e-05,
+      "loss": 0.9209,
+      "step": 189270
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.562158209819185e-05,
+      "loss": 0.9016,
+      "step": 189280
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.562082044876384e-05,
+      "loss": 0.8993,
+      "step": 189290
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.5620058799335844e-05,
+      "loss": 0.8959,
+      "step": 189300
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.561929714990784e-05,
+      "loss": 0.7641,
+      "step": 189310
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.561853550047984e-05,
+      "loss": 0.855,
+      "step": 189320
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.5617773851051837e-05,
+      "loss": 0.947,
+      "step": 189330
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561701220162384e-05,
+      "loss": 0.8459,
+      "step": 189340
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561625055219583e-05,
+      "loss": 0.9542,
+      "step": 189350
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.5615488902767836e-05,
+      "loss": 0.987,
+      "step": 189360
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561472725333983e-05,
+      "loss": 0.934,
+      "step": 189370
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561396560391183e-05,
+      "loss": 0.8764,
+      "step": 189380
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561320395448383e-05,
+      "loss": 0.9259,
+      "step": 189390
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561244230505583e-05,
+      "loss": 0.924,
+      "step": 189400
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561168065562783e-05,
+      "loss": 0.8438,
+      "step": 189410
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561091900619983e-05,
+      "loss": 0.9174,
+      "step": 189420
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.561015735677183e-05,
+      "loss": 0.8449,
+      "step": 189430
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560939570734382e-05,
+      "loss": 0.8788,
+      "step": 189440
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.5608634057915826e-05,
+      "loss": 1.021,
+      "step": 189450
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.5607872408487824e-05,
+      "loss": 0.856,
+      "step": 189460
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560711075905982e-05,
+      "loss": 0.9844,
+      "step": 189470
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560634910963182e-05,
+      "loss": 0.8743,
+      "step": 189480
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560558746020382e-05,
+      "loss": 0.9808,
+      "step": 189490
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.5604825810775814e-05,
+      "loss": 0.8976,
+      "step": 189500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560406416134782e-05,
+      "loss": 0.9395,
+      "step": 189510
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.5603302511919816e-05,
+      "loss": 0.866,
+      "step": 189520
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560254086249181e-05,
+      "loss": 0.8906,
+      "step": 189530
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.560177921306381e-05,
+      "loss": 0.8761,
+      "step": 189540
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5601017563635815e-05,
+      "loss": 0.9155,
+      "step": 189550
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5600255914207805e-05,
+      "loss": 0.9177,
+      "step": 189560
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.559949426477981e-05,
+      "loss": 0.8044,
+      "step": 189570
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.559873261535181e-05,
+      "loss": 0.8377,
+      "step": 189580
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5597970965923805e-05,
+      "loss": 0.8659,
+      "step": 189590
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.55972093164958e-05,
+      "loss": 0.8058,
+      "step": 189600
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5596447667067806e-05,
+      "loss": 0.8995,
+      "step": 189610
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5595686017639804e-05,
+      "loss": 1.0027,
+      "step": 189620
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.55949243682118e-05,
+      "loss": 0.8558,
+      "step": 189630
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5594162718783805e-05,
+      "loss": 0.7897,
+      "step": 189640
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5593401069355796e-05,
+      "loss": 0.9427,
+      "step": 189650
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.55926394199278e-05,
+      "loss": 0.8388,
+      "step": 189660
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.55918777704998e-05,
+      "loss": 0.9517,
+      "step": 189670
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5591116121071795e-05,
+      "loss": 0.9406,
+      "step": 189680
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.559035447164379e-05,
+      "loss": 0.83,
+      "step": 189690
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.55895928222158e-05,
+      "loss": 0.8276,
+      "step": 189700
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.558883117278779e-05,
+      "loss": 0.883,
+      "step": 189710
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.558806952335979e-05,
+      "loss": 0.8381,
+      "step": 189720
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.558730787393179e-05,
+      "loss": 0.9076,
+      "step": 189730
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.558654622450379e-05,
+      "loss": 0.8839,
+      "step": 189740
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.5585784575075784e-05,
+      "loss": 1.032,
+      "step": 189750
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.558502292564778e-05,
+      "loss": 0.8234,
+      "step": 189760
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.558426127621978e-05,
+      "loss": 0.7669,
+      "step": 189770
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5583499626791784e-05,
+      "loss": 0.8929,
+      "step": 189780
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.558273797736378e-05,
+      "loss": 0.8116,
+      "step": 189790
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.558197632793578e-05,
+      "loss": 0.8141,
+      "step": 189800
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5581214678507776e-05,
+      "loss": 0.9288,
+      "step": 189810
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5580453029079773e-05,
+      "loss": 0.8804,
+      "step": 189820
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557969137965178e-05,
+      "loss": 0.8692,
+      "step": 189830
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5578929730223775e-05,
+      "loss": 0.8385,
+      "step": 189840
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557816808079577e-05,
+      "loss": 0.9136,
+      "step": 189850
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557740643136777e-05,
+      "loss": 0.8715,
+      "step": 189860
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5576644781939774e-05,
+      "loss": 0.8807,
+      "step": 189870
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5575883132511765e-05,
+      "loss": 0.897,
+      "step": 189880
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557512148308377e-05,
+      "loss": 0.925,
+      "step": 189890
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557435983365577e-05,
+      "loss": 0.8363,
+      "step": 189900
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5573598184227764e-05,
+      "loss": 0.9089,
+      "step": 189910
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557283653479976e-05,
+      "loss": 0.9937,
+      "step": 189920
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5572074885371766e-05,
+      "loss": 1.0341,
+      "step": 189930
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5571313235943757e-05,
+      "loss": 0.9094,
+      "step": 189940
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.557055158651576e-05,
+      "loss": 0.9372,
+      "step": 189950
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.556978993708776e-05,
+      "loss": 0.8732,
+      "step": 189960
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.5569028287659756e-05,
+      "loss": 0.9662,
+      "step": 189970
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.556826663823175e-05,
+      "loss": 0.866,
+      "step": 189980
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556750498880376e-05,
+      "loss": 0.8891,
+      "step": 189990
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.5566743339375755e-05,
+      "loss": 0.9601,
+      "step": 190000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556598168994775e-05,
+      "loss": 0.9279,
+      "step": 190010
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556522004051976e-05,
+      "loss": 0.9689,
+      "step": 190020
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556445839109175e-05,
+      "loss": 0.9283,
+      "step": 190030
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556369674166375e-05,
+      "loss": 0.8028,
+      "step": 190040
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556293509223575e-05,
+      "loss": 0.7881,
+      "step": 190050
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.5562173442807747e-05,
+      "loss": 0.9051,
+      "step": 190060
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.5561411793379744e-05,
+      "loss": 0.9911,
+      "step": 190070
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.556065014395175e-05,
+      "loss": 0.8547,
+      "step": 190080
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555988849452374e-05,
+      "loss": 0.8424,
+      "step": 190090
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555912684509574e-05,
+      "loss": 0.8381,
+      "step": 190100
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555836519566774e-05,
+      "loss": 0.8798,
+      "step": 190110
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555760354623974e-05,
+      "loss": 0.7892,
+      "step": 190120
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.5556841896811736e-05,
+      "loss": 0.9035,
+      "step": 190130
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555608024738374e-05,
+      "loss": 0.8963,
+      "step": 190140
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555531859795573e-05,
+      "loss": 0.8168,
+      "step": 190150
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.5554556948527735e-05,
+      "loss": 0.9014,
+      "step": 190160
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555379529909973e-05,
+      "loss": 0.7477,
+      "step": 190170
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555303364967173e-05,
+      "loss": 0.8468,
+      "step": 190180
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.555227200024373e-05,
+      "loss": 0.7811,
+      "step": 190190
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.555151035081573e-05,
+      "loss": 0.9529,
+      "step": 190200
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.555074870138773e-05,
+      "loss": 0.814,
+      "step": 190210
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5549987051959726e-05,
+      "loss": 0.8462,
+      "step": 190220
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5549225402531724e-05,
+      "loss": 0.9077,
+      "step": 190230
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554846375310372e-05,
+      "loss": 0.9321,
+      "step": 190240
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5547702103675725e-05,
+      "loss": 0.9001,
+      "step": 190250
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5546940454247716e-05,
+      "loss": 0.8127,
+      "step": 190260
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554617880481972e-05,
+      "loss": 0.8993,
+      "step": 190270
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554541715539172e-05,
+      "loss": 0.8933,
+      "step": 190280
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5544655505963715e-05,
+      "loss": 0.8396,
+      "step": 190290
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554389385653571e-05,
+      "loss": 0.8466,
+      "step": 190300
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554313220710772e-05,
+      "loss": 0.9481,
+      "step": 190310
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554237055767971e-05,
+      "loss": 0.8755,
+      "step": 190320
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554160890825171e-05,
+      "loss": 0.8835,
+      "step": 190330
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554084725882371e-05,
+      "loss": 0.9204,
+      "step": 190340
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.554008560939571e-05,
+      "loss": 0.9121,
+      "step": 190350
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5539323959967704e-05,
+      "loss": 0.8872,
+      "step": 190360
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.553856231053971e-05,
+      "loss": 1.0621,
+      "step": 190370
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5537800661111706e-05,
+      "loss": 0.7604,
+      "step": 190380
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.5537039011683704e-05,
+      "loss": 0.9732,
+      "step": 190390
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.55362773622557e-05,
+      "loss": 0.8262,
+      "step": 190400
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.55355157128277e-05,
+      "loss": 0.8009,
+      "step": 190410
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.55347540633997e-05,
+      "loss": 0.9404,
+      "step": 190420
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.55339924139717e-05,
+      "loss": 0.9139,
+      "step": 190430
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.55332307645437e-05,
+      "loss": 0.9034,
+      "step": 190440
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.5532469115115695e-05,
+      "loss": 0.7959,
+      "step": 190450
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.55317074656877e-05,
+      "loss": 0.8964,
+      "step": 190460
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.553094581625969e-05,
+      "loss": 1.022,
+      "step": 190470
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.5530184166831694e-05,
+      "loss": 0.8597,
+      "step": 190480
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552942251740369e-05,
+      "loss": 0.8122,
+      "step": 190490
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552866086797569e-05,
+      "loss": 0.8731,
+      "step": 190500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552789921854769e-05,
+      "loss": 0.8746,
+      "step": 190510
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552713756911969e-05,
+      "loss": 0.9016,
+      "step": 190520
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552637591969168e-05,
+      "loss": 0.9281,
+      "step": 190530
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.5525614270263686e-05,
+      "loss": 0.9162,
+      "step": 190540
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552485262083568e-05,
+      "loss": 0.8201,
+      "step": 190550
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552409097140768e-05,
+      "loss": 0.8954,
+      "step": 190560
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552332932197968e-05,
+      "loss": 0.8952,
+      "step": 190570
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552256767255168e-05,
+      "loss": 0.8932,
+      "step": 190580
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552180602312368e-05,
+      "loss": 0.9584,
+      "step": 190590
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552104437369568e-05,
+      "loss": 0.9082,
+      "step": 190600
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.552028272426768e-05,
+      "loss": 0.9166,
+      "step": 190610
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.551952107483967e-05,
+      "loss": 0.8613,
+      "step": 190620
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.551875942541168e-05,
+      "loss": 0.8671,
+      "step": 190630
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5517997775983674e-05,
+      "loss": 0.9805,
+      "step": 190640
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551723612655567e-05,
+      "loss": 0.8583,
+      "step": 190650
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551647447712767e-05,
+      "loss": 0.9571,
+      "step": 190660
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5515712827699667e-05,
+      "loss": 0.9136,
+      "step": 190670
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5514951178271664e-05,
+      "loss": 0.8583,
+      "step": 190680
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551418952884367e-05,
+      "loss": 0.8541,
+      "step": 190690
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551342787941566e-05,
+      "loss": 0.8441,
+      "step": 190700
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551266622998766e-05,
+      "loss": 0.8096,
+      "step": 190710
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551190458055966e-05,
+      "loss": 0.8298,
+      "step": 190720
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.551114293113166e-05,
+      "loss": 0.9155,
+      "step": 190730
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5510381281703656e-05,
+      "loss": 0.8308,
+      "step": 190740
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550961963227566e-05,
+      "loss": 1.0408,
+      "step": 190750
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550885798284766e-05,
+      "loss": 0.8954,
+      "step": 190760
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5508096333419655e-05,
+      "loss": 0.8357,
+      "step": 190770
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550733468399165e-05,
+      "loss": 0.8983,
+      "step": 190780
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550657303456365e-05,
+      "loss": 0.9393,
+      "step": 190790
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5505811385135654e-05,
+      "loss": 0.9268,
+      "step": 190800
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550504973570765e-05,
+      "loss": 0.9777,
+      "step": 190810
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550428808627965e-05,
+      "loss": 0.8877,
+      "step": 190820
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.5503526436851646e-05,
+      "loss": 0.9445,
+      "step": 190830
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550276478742365e-05,
+      "loss": 0.9096,
+      "step": 190840
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.550200313799564e-05,
+      "loss": 0.8389,
+      "step": 190850
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5501241488567645e-05,
+      "loss": 0.8093,
+      "step": 190860
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.550047983913964e-05,
+      "loss": 0.9157,
+      "step": 190870
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549971818971164e-05,
+      "loss": 0.9382,
+      "step": 190880
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549895654028364e-05,
+      "loss": 0.8614,
+      "step": 190890
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549819489085564e-05,
+      "loss": 0.9118,
+      "step": 190900
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549743324142763e-05,
+      "loss": 0.9619,
+      "step": 190910
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549667159199964e-05,
+      "loss": 0.8373,
+      "step": 190920
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5495909942571635e-05,
+      "loss": 0.8729,
+      "step": 190930
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549514829314363e-05,
+      "loss": 0.7838,
+      "step": 190940
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549438664371563e-05,
+      "loss": 0.8873,
+      "step": 190950
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5493624994287634e-05,
+      "loss": 0.9226,
+      "step": 190960
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549286334485963e-05,
+      "loss": 0.9737,
+      "step": 190970
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.549210169543163e-05,
+      "loss": 0.9397,
+      "step": 190980
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5491340046003626e-05,
+      "loss": 0.9255,
+      "step": 190990
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5490578396575624e-05,
+      "loss": 0.856,
+      "step": 191000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.548981674714763e-05,
+      "loss": 0.9764,
+      "step": 191010
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5489055097719625e-05,
+      "loss": 0.8105,
+      "step": 191020
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.548829344829162e-05,
+      "loss": 0.9386,
+      "step": 191030
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.548753179886362e-05,
+      "loss": 0.8891,
+      "step": 191040
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5486770149435624e-05,
+      "loss": 0.9386,
+      "step": 191050
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.5486008500007615e-05,
+      "loss": 1.1088,
+      "step": 191060
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.548524685057962e-05,
+      "loss": 0.8474,
+      "step": 191070
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.548448520115162e-05,
+      "loss": 0.8694,
+      "step": 191080
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.5483723551723614e-05,
+      "loss": 0.8299,
+      "step": 191090
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.548296190229561e-05,
+      "loss": 0.9141,
+      "step": 191100
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.5482200252867616e-05,
+      "loss": 0.9408,
+      "step": 191110
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.548143860343961e-05,
+      "loss": 0.955,
+      "step": 191120
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.548067695401161e-05,
+      "loss": 0.8974,
+      "step": 191130
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.547991530458361e-05,
+      "loss": 1.024,
+      "step": 191140
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.5479153655155606e-05,
+      "loss": 0.9093,
+      "step": 191150
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54783920057276e-05,
+      "loss": 0.9588,
+      "step": 191160
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54776303562996e-05,
+      "loss": 0.8765,
+      "step": 191170
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.5476868706871605e-05,
+      "loss": 0.9167,
+      "step": 191180
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54761070574436e-05,
+      "loss": 0.8888,
+      "step": 191190
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54753454080156e-05,
+      "loss": 0.9461,
+      "step": 191200
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54745837585876e-05,
+      "loss": 0.8561,
+      "step": 191210
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54738221091596e-05,
+      "loss": 0.916,
+      "step": 191220
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.547306045973159e-05,
+      "loss": 0.8788,
+      "step": 191230
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.54722988103036e-05,
+      "loss": 0.8502,
+      "step": 191240
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.5471537160875594e-05,
+      "loss": 1.0102,
+      "step": 191250
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.547077551144759e-05,
+      "loss": 0.9163,
+      "step": 191260
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.547001386201959e-05,
+      "loss": 0.9151,
+      "step": 191270
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.546925221259159e-05,
+      "loss": 0.9378,
+      "step": 191280
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.5468490563163584e-05,
+      "loss": 0.9262,
+      "step": 191290
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546772891373559e-05,
+      "loss": 0.908,
+      "step": 191300
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5466967264307586e-05,
+      "loss": 0.878,
+      "step": 191310
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546620561487958e-05,
+      "loss": 0.9117,
+      "step": 191320
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546544396545158e-05,
+      "loss": 1.018,
+      "step": 191330
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5464682316023585e-05,
+      "loss": 1.0594,
+      "step": 191340
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546392066659558e-05,
+      "loss": 0.9476,
+      "step": 191350
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546315901716758e-05,
+      "loss": 0.8924,
+      "step": 191360
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546239736773958e-05,
+      "loss": 0.9778,
+      "step": 191370
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5461635718311575e-05,
+      "loss": 0.7528,
+      "step": 191380
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.546087406888358e-05,
+      "loss": 0.9768,
+      "step": 191390
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5460112419455576e-05,
+      "loss": 0.8328,
+      "step": 191400
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5459350770027574e-05,
+      "loss": 0.9175,
+      "step": 191410
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545858912059957e-05,
+      "loss": 0.8475,
+      "step": 191420
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5457827471171576e-05,
+      "loss": 1.0059,
+      "step": 191430
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5457065821743566e-05,
+      "loss": 0.8253,
+      "step": 191440
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545630417231557e-05,
+      "loss": 0.8733,
+      "step": 191450
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545554252288757e-05,
+      "loss": 0.8614,
+      "step": 191460
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.5454780873459566e-05,
+      "loss": 0.9058,
+      "step": 191470
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545401922403156e-05,
+      "loss": 0.8847,
+      "step": 191480
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545325757460357e-05,
+      "loss": 0.8814,
+      "step": 191490
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545249592517556e-05,
+      "loss": 0.7734,
+      "step": 191500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.545173427574756e-05,
+      "loss": 0.7439,
+      "step": 191510
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.545097262631956e-05,
+      "loss": 0.7873,
+      "step": 191520
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.545021097689156e-05,
+      "loss": 0.9382,
+      "step": 191530
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5449449327463555e-05,
+      "loss": 0.965,
+      "step": 191540
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544868767803556e-05,
+      "loss": 0.7837,
+      "step": 191550
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5447926028607556e-05,
+      "loss": 0.8923,
+      "step": 191560
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5447164379179554e-05,
+      "loss": 0.8899,
+      "step": 191570
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544640272975155e-05,
+      "loss": 0.9845,
+      "step": 191580
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544564108032355e-05,
+      "loss": 0.9242,
+      "step": 191590
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544487943089555e-05,
+      "loss": 0.7759,
+      "step": 191600
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5444117781467544e-05,
+      "loss": 0.9491,
+      "step": 191610
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544335613203955e-05,
+      "loss": 0.9,
+      "step": 191620
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5442594482611545e-05,
+      "loss": 0.8776,
+      "step": 191630
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544183283318354e-05,
+      "loss": 0.8963,
+      "step": 191640
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.544107118375554e-05,
+      "loss": 0.816,
+      "step": 191650
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5440309534327544e-05,
+      "loss": 0.9589,
+      "step": 191660
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5439547884899535e-05,
+      "loss": 1.0385,
+      "step": 191670
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.543878623547154e-05,
+      "loss": 0.8977,
+      "step": 191680
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.543802458604354e-05,
+      "loss": 0.9347,
+      "step": 191690
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5437262936615534e-05,
+      "loss": 0.9019,
+      "step": 191700
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.543650128718753e-05,
+      "loss": 0.8614,
+      "step": 191710
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5435739637759536e-05,
+      "loss": 0.9612,
+      "step": 191720
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.5434977988331534e-05,
+      "loss": 0.8221,
+      "step": 191730
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.543421633890353e-05,
+      "loss": 0.8444,
+      "step": 191740
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.543345468947553e-05,
+      "loss": 0.9181,
+      "step": 191750
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.5432693040047526e-05,
+      "loss": 0.8217,
+      "step": 191760
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.543193139061953e-05,
+      "loss": 0.8196,
+      "step": 191770
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.543116974119153e-05,
+      "loss": 0.9641,
+      "step": 191780
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.5430408091763525e-05,
+      "loss": 0.8734,
+      "step": 191790
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542964644233552e-05,
+      "loss": 0.9639,
+      "step": 191800
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542888479290753e-05,
+      "loss": 0.9174,
+      "step": 191810
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542812314347952e-05,
+      "loss": 0.8915,
+      "step": 191820
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542736149405152e-05,
+      "loss": 0.8943,
+      "step": 191830
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542659984462352e-05,
+      "loss": 0.8638,
+      "step": 191840
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542583819519552e-05,
+      "loss": 0.9162,
+      "step": 191850
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.5425076545767514e-05,
+      "loss": 0.8734,
+      "step": 191860
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542431489633952e-05,
+      "loss": 1.028,
+      "step": 191870
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542355324691151e-05,
+      "loss": 0.8376,
+      "step": 191880
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542279159748351e-05,
+      "loss": 0.9118,
+      "step": 191890
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542202994805551e-05,
+      "loss": 0.8832,
+      "step": 191900
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.542126829862751e-05,
+      "loss": 0.9007,
+      "step": 191910
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.5420506649199506e-05,
+      "loss": 0.8467,
+      "step": 191920
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.541974499977151e-05,
+      "loss": 0.9009,
+      "step": 191930
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.541898335034351e-05,
+      "loss": 0.9511,
+      "step": 191940
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.5418221700915505e-05,
+      "loss": 0.8164,
+      "step": 191950
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.54174600514875e-05,
+      "loss": 0.8252,
+      "step": 191960
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.54166984020595e-05,
+      "loss": 0.9632,
+      "step": 191970
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5415936752631504e-05,
+      "loss": 0.901,
+      "step": 191980
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.54151751032035e-05,
+      "loss": 0.8942,
+      "step": 191990
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.54144134537755e-05,
+      "loss": 0.951,
+      "step": 192000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5413651804347496e-05,
+      "loss": 0.9501,
+      "step": 192010
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.54128901549195e-05,
+      "loss": 0.8488,
+      "step": 192020
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.541212850549149e-05,
+      "loss": 1.0937,
+      "step": 192030
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5411366856063496e-05,
+      "loss": 0.8498,
+      "step": 192040
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5410605206635486e-05,
+      "loss": 0.9148,
+      "step": 192050
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540984355720749e-05,
+      "loss": 0.9486,
+      "step": 192060
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540908190777949e-05,
+      "loss": 0.8112,
+      "step": 192070
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5408320258351486e-05,
+      "loss": 0.8812,
+      "step": 192080
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540755860892348e-05,
+      "loss": 0.755,
+      "step": 192090
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540679695949549e-05,
+      "loss": 0.8249,
+      "step": 192100
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5406035310067485e-05,
+      "loss": 0.9781,
+      "step": 192110
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540527366063948e-05,
+      "loss": 0.888,
+      "step": 192120
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540451201121148e-05,
+      "loss": 0.7244,
+      "step": 192130
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540375036178348e-05,
+      "loss": 0.9624,
+      "step": 192140
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540298871235548e-05,
+      "loss": 0.9062,
+      "step": 192150
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.540222706292748e-05,
+      "loss": 0.849,
+      "step": 192160
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.5401465413499476e-05,
+      "loss": 0.8382,
+      "step": 192170
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.5400703764071474e-05,
+      "loss": 0.9201,
+      "step": 192180
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539994211464348e-05,
+      "loss": 0.9574,
+      "step": 192190
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539918046521547e-05,
+      "loss": 0.936,
+      "step": 192200
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539841881578747e-05,
+      "loss": 0.8959,
+      "step": 192210
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539765716635947e-05,
+      "loss": 0.9964,
+      "step": 192220
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539689551693147e-05,
+      "loss": 0.8188,
+      "step": 192230
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.5396133867503465e-05,
+      "loss": 0.8927,
+      "step": 192240
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539537221807547e-05,
+      "loss": 0.7613,
+      "step": 192250
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539461056864746e-05,
+      "loss": 0.7703,
+      "step": 192260
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.5393848919219464e-05,
+      "loss": 0.8939,
+      "step": 192270
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539308726979146e-05,
+      "loss": 0.8747,
+      "step": 192280
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539232562036346e-05,
+      "loss": 0.9597,
+      "step": 192290
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539156397093546e-05,
+      "loss": 0.9644,
+      "step": 192300
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539080232150746e-05,
+      "loss": 0.7904,
+      "step": 192310
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.539004067207946e-05,
+      "loss": 0.8151,
+      "step": 192320
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.5389279022651456e-05,
+      "loss": 0.9526,
+      "step": 192330
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.5388517373223454e-05,
+      "loss": 0.8164,
+      "step": 192340
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.538775572379545e-05,
+      "loss": 0.8979,
+      "step": 192350
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.5386994074367455e-05,
+      "loss": 0.9693,
+      "step": 192360
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.538623242493945e-05,
+      "loss": 0.8085,
+      "step": 192370
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.538547077551145e-05,
+      "loss": 0.8207,
+      "step": 192380
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.538470912608345e-05,
+      "loss": 0.7928,
+      "step": 192390
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.538394747665545e-05,
+      "loss": 0.9338,
+      "step": 192400
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.538318582722744e-05,
+      "loss": 0.9602,
+      "step": 192410
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.538242417779945e-05,
+      "loss": 0.8487,
+      "step": 192420
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.5381662528371444e-05,
+      "loss": 0.9669,
+      "step": 192430
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.538090087894344e-05,
+      "loss": 0.8976,
+      "step": 192440
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.538013922951544e-05,
+      "loss": 0.8863,
+      "step": 192450
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.5379377580087443e-05,
+      "loss": 0.9383,
+      "step": 192460
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.5378615930659434e-05,
+      "loss": 0.8392,
+      "step": 192470
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537785428123144e-05,
+      "loss": 0.8468,
+      "step": 192480
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.5377092631803436e-05,
+      "loss": 0.8308,
+      "step": 192490
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537633098237543e-05,
+      "loss": 0.8824,
+      "step": 192500
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537556933294743e-05,
+      "loss": 0.8737,
+      "step": 192510
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537480768351943e-05,
+      "loss": 0.8577,
+      "step": 192520
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537404603409143e-05,
+      "loss": 0.8545,
+      "step": 192530
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537328438466343e-05,
+      "loss": 0.8917,
+      "step": 192540
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537252273523543e-05,
+      "loss": 0.9527,
+      "step": 192550
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.5371761085807425e-05,
+      "loss": 0.8512,
+      "step": 192560
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537099943637943e-05,
+      "loss": 0.958,
+      "step": 192570
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.537023778695142e-05,
+      "loss": 0.9158,
+      "step": 192580
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.5369476137523424e-05,
+      "loss": 0.9006,
+      "step": 192590
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.536871448809542e-05,
+      "loss": 0.8812,
+      "step": 192600
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536795283866742e-05,
+      "loss": 0.9223,
+      "step": 192610
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5367191189239416e-05,
+      "loss": 0.9407,
+      "step": 192620
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536642953981142e-05,
+      "loss": 0.931,
+      "step": 192630
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536566789038341e-05,
+      "loss": 0.9176,
+      "step": 192640
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5364906240955416e-05,
+      "loss": 0.8006,
+      "step": 192650
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536414459152741e-05,
+      "loss": 0.8958,
+      "step": 192660
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536338294209941e-05,
+      "loss": 0.8991,
+      "step": 192670
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536262129267141e-05,
+      "loss": 0.905,
+      "step": 192680
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536185964324341e-05,
+      "loss": 0.9897,
+      "step": 192690
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.53610979938154e-05,
+      "loss": 1.0243,
+      "step": 192700
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.536033634438741e-05,
+      "loss": 0.857,
+      "step": 192710
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5359574694959405e-05,
+      "loss": 1.0026,
+      "step": 192720
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.53588130455314e-05,
+      "loss": 0.9382,
+      "step": 192730
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5358051396103406e-05,
+      "loss": 0.8789,
+      "step": 192740
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5357289746675404e-05,
+      "loss": 0.8488,
+      "step": 192750
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.53565280972474e-05,
+      "loss": 0.9925,
+      "step": 192760
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.53557664478194e-05,
+      "loss": 0.8852,
+      "step": 192770
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.53550047983914e-05,
+      "loss": 0.855,
+      "step": 192780
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5354243148963394e-05,
+      "loss": 0.9034,
+      "step": 192790
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.53534814995354e-05,
+      "loss": 0.9806,
+      "step": 192800
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.5352719850107395e-05,
+      "loss": 0.8851,
+      "step": 192810
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.535195820067939e-05,
+      "loss": 0.9082,
+      "step": 192820
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.535119655125139e-05,
+      "loss": 0.8586,
+      "step": 192830
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5350434901823395e-05,
+      "loss": 0.8566,
+      "step": 192840
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5349673252395385e-05,
+      "loss": 0.8733,
+      "step": 192850
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534891160296739e-05,
+      "loss": 0.7241,
+      "step": 192860
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534814995353939e-05,
+      "loss": 0.8411,
+      "step": 192870
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5347388304111384e-05,
+      "loss": 0.7873,
+      "step": 192880
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534662665468338e-05,
+      "loss": 0.9226,
+      "step": 192890
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5345865005255386e-05,
+      "loss": 0.8805,
+      "step": 192900
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5345103355827384e-05,
+      "loss": 0.8524,
+      "step": 192910
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534434170639938e-05,
+      "loss": 0.8879,
+      "step": 192920
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534358005697138e-05,
+      "loss": 0.9809,
+      "step": 192930
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5342818407543376e-05,
+      "loss": 0.8827,
+      "step": 192940
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534205675811538e-05,
+      "loss": 0.7995,
+      "step": 192950
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.534129510868738e-05,
+      "loss": 0.8555,
+      "step": 192960
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5340533459259375e-05,
+      "loss": 0.9156,
+      "step": 192970
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.533977180983137e-05,
+      "loss": 0.8445,
+      "step": 192980
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.533901016040337e-05,
+      "loss": 0.9109,
+      "step": 192990
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.533824851097537e-05,
+      "loss": 0.9314,
+      "step": 193000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.533748686154737e-05,
+      "loss": 0.8425,
+      "step": 193010
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.533672521211936e-05,
+      "loss": 0.8836,
+      "step": 193020
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.533596356269137e-05,
+      "loss": 0.8329,
+      "step": 193030
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5335201913263364e-05,
+      "loss": 0.8823,
+      "step": 193040
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.533444026383536e-05,
+      "loss": 0.9579,
+      "step": 193050
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.533367861440736e-05,
+      "loss": 0.9017,
+      "step": 193060
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5332916964979363e-05,
+      "loss": 1.0114,
+      "step": 193070
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5332155315551354e-05,
+      "loss": 0.8294,
+      "step": 193080
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.533139366612336e-05,
+      "loss": 0.9369,
+      "step": 193090
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5330632016695356e-05,
+      "loss": 0.9337,
+      "step": 193100
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532987036726735e-05,
+      "loss": 0.7799,
+      "step": 193110
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532910871783936e-05,
+      "loss": 0.8983,
+      "step": 193120
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5328347068411355e-05,
+      "loss": 0.8915,
+      "step": 193130
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532758541898335e-05,
+      "loss": 0.8928,
+      "step": 193140
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532682376955535e-05,
+      "loss": 0.8758,
+      "step": 193150
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5326062120127354e-05,
+      "loss": 1.0047,
+      "step": 193160
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5325300470699345e-05,
+      "loss": 0.7899,
+      "step": 193170
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532453882127135e-05,
+      "loss": 0.8999,
+      "step": 193180
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532377717184335e-05,
+      "loss": 0.9402,
+      "step": 193190
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5323015522415344e-05,
+      "loss": 0.9065,
+      "step": 193200
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.532225387298734e-05,
+      "loss": 0.9965,
+      "step": 193210
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5321492223559346e-05,
+      "loss": 0.8314,
+      "step": 193220
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5320730574131336e-05,
+      "loss": 0.8921,
+      "step": 193230
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.531996892470334e-05,
+      "loss": 0.8683,
+      "step": 193240
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.531920727527534e-05,
+      "loss": 0.8119,
+      "step": 193250
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5318445625847336e-05,
+      "loss": 0.8901,
+      "step": 193260
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531768397641933e-05,
+      "loss": 0.9345,
+      "step": 193270
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531692232699134e-05,
+      "loss": 0.9069,
+      "step": 193280
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531616067756333e-05,
+      "loss": 0.8991,
+      "step": 193290
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531539902813533e-05,
+      "loss": 0.8775,
+      "step": 193300
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531463737870733e-05,
+      "loss": 0.8598,
+      "step": 193310
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531387572927933e-05,
+      "loss": 0.9413,
+      "step": 193320
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531311407985133e-05,
+      "loss": 1.0272,
+      "step": 193330
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531235243042333e-05,
+      "loss": 0.9653,
+      "step": 193340
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5311590780995326e-05,
+      "loss": 0.8721,
+      "step": 193350
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5310829131567324e-05,
+      "loss": 0.961,
+      "step": 193360
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.531006748213933e-05,
+      "loss": 0.8636,
+      "step": 193370
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530930583271132e-05,
+      "loss": 0.9121,
+      "step": 193380
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530854418328332e-05,
+      "loss": 0.928,
+      "step": 193390
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530778253385532e-05,
+      "loss": 0.7844,
+      "step": 193400
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530702088442732e-05,
+      "loss": 0.9804,
+      "step": 193410
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5306259234999315e-05,
+      "loss": 0.999,
+      "step": 193420
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530549758557132e-05,
+      "loss": 0.9557,
+      "step": 193430
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530473593614331e-05,
+      "loss": 0.9334,
+      "step": 193440
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5303974286715315e-05,
+      "loss": 0.8215,
+      "step": 193450
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5303212637287305e-05,
+      "loss": 0.9075,
+      "step": 193460
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530245098785931e-05,
+      "loss": 0.942,
+      "step": 193470
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530168933843131e-05,
+      "loss": 0.9543,
+      "step": 193480
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5300927689003305e-05,
+      "loss": 1.0107,
+      "step": 193490
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.53001660395753e-05,
+      "loss": 0.834,
+      "step": 193500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5299404390147306e-05,
+      "loss": 0.9351,
+      "step": 193510
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5298642740719304e-05,
+      "loss": 0.8244,
+      "step": 193520
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.52978810912913e-05,
+      "loss": 0.8903,
+      "step": 193530
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5297119441863305e-05,
+      "loss": 0.8728,
+      "step": 193540
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5296357792435296e-05,
+      "loss": 1.1136,
+      "step": 193550
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.52955961430073e-05,
+      "loss": 0.8871,
+      "step": 193560
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.52948344935793e-05,
+      "loss": 0.8792,
+      "step": 193570
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5294072844151295e-05,
+      "loss": 0.9502,
+      "step": 193580
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.529331119472329e-05,
+      "loss": 0.94,
+      "step": 193590
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.52925495452953e-05,
+      "loss": 0.9382,
+      "step": 193600
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.529178789586729e-05,
+      "loss": 0.9412,
+      "step": 193610
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.529102624643929e-05,
+      "loss": 0.7919,
+      "step": 193620
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.529026459701129e-05,
+      "loss": 0.9583,
+      "step": 193630
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.528950294758329e-05,
+      "loss": 0.8645,
+      "step": 193640
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5288741298155284e-05,
+      "loss": 0.8595,
+      "step": 193650
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.528797964872729e-05,
+      "loss": 0.8864,
+      "step": 193660
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.528721799929928e-05,
+      "loss": 0.9334,
+      "step": 193670
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.5286456349871283e-05,
+      "loss": 0.9035,
+      "step": 193680
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.528569470044328e-05,
+      "loss": 0.9557,
+      "step": 193690
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.528493305101528e-05,
+      "loss": 0.8515,
+      "step": 193700
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.528417140158728e-05,
+      "loss": 0.9671,
+      "step": 193710
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.528340975215928e-05,
+      "loss": 0.8471,
+      "step": 193720
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.528264810273128e-05,
+      "loss": 0.8766,
+      "step": 193730
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.5281886453303275e-05,
+      "loss": 1.0419,
+      "step": 193740
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.528112480387528e-05,
+      "loss": 0.9115,
+      "step": 193750
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.528036315444727e-05,
+      "loss": 0.9698,
+      "step": 193760
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.5279601505019274e-05,
+      "loss": 0.8826,
+      "step": 193770
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527883985559127e-05,
+      "loss": 0.9472,
+      "step": 193780
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527807820616327e-05,
+      "loss": 1.0087,
+      "step": 193790
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527731655673527e-05,
+      "loss": 0.8642,
+      "step": 193800
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527655490730727e-05,
+      "loss": 0.967,
+      "step": 193810
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527579325787926e-05,
+      "loss": 0.8179,
+      "step": 193820
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.5275031608451266e-05,
+      "loss": 0.9135,
+      "step": 193830
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527426995902326e-05,
+      "loss": 0.9608,
+      "step": 193840
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527350830959526e-05,
+      "loss": 0.8333,
+      "step": 193850
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527274666016726e-05,
+      "loss": 0.8167,
+      "step": 193860
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527198501073926e-05,
+      "loss": 0.9901,
+      "step": 193870
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527122336131125e-05,
+      "loss": 0.9504,
+      "step": 193880
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.527046171188326e-05,
+      "loss": 0.9363,
+      "step": 193890
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.5269700062455255e-05,
+      "loss": 0.9801,
+      "step": 193900
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.526893841302725e-05,
+      "loss": 0.9038,
+      "step": 193910
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.5268176763599257e-05,
+      "loss": 0.9101,
+      "step": 193920
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526741511417125e-05,
+      "loss": 0.8409,
+      "step": 193930
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526665346474325e-05,
+      "loss": 0.9329,
+      "step": 193940
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526589181531525e-05,
+      "loss": 0.8582,
+      "step": 193950
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.5265130165887246e-05,
+      "loss": 0.8968,
+      "step": 193960
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.5264368516459244e-05,
+      "loss": 0.961,
+      "step": 193970
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526360686703125e-05,
+      "loss": 0.8796,
+      "step": 193980
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526284521760324e-05,
+      "loss": 0.875,
+      "step": 193990
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526208356817524e-05,
+      "loss": 0.8481,
+      "step": 194000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526132191874724e-05,
+      "loss": 0.8596,
+      "step": 194010
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.526056026931924e-05,
+      "loss": 0.8265,
+      "step": 194020
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.5259798619891235e-05,
+      "loss": 0.9599,
+      "step": 194030
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525903697046324e-05,
+      "loss": 0.8905,
+      "step": 194040
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525827532103523e-05,
+      "loss": 0.9389,
+      "step": 194050
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.5257513671607235e-05,
+      "loss": 0.8732,
+      "step": 194060
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525675202217923e-05,
+      "loss": 0.7447,
+      "step": 194070
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525599037275123e-05,
+      "loss": 0.7819,
+      "step": 194080
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525522872332323e-05,
+      "loss": 0.7899,
+      "step": 194090
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525446707389523e-05,
+      "loss": 0.9029,
+      "step": 194100
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525370542446723e-05,
+      "loss": 1.0383,
+      "step": 194110
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.5252943775039226e-05,
+      "loss": 0.927,
+      "step": 194120
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525218212561123e-05,
+      "loss": 0.8576,
+      "step": 194130
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.525142047618322e-05,
+      "loss": 1.0363,
+      "step": 194140
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5250658826755225e-05,
+      "loss": 0.8113,
+      "step": 194150
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524989717732722e-05,
+      "loss": 0.8897,
+      "step": 194160
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524913552789922e-05,
+      "loss": 0.9403,
+      "step": 194170
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524837387847122e-05,
+      "loss": 0.8261,
+      "step": 194180
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524761222904322e-05,
+      "loss": 0.8545,
+      "step": 194190
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524685057961521e-05,
+      "loss": 0.8732,
+      "step": 194200
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524608893018722e-05,
+      "loss": 0.9564,
+      "step": 194210
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5245327280759214e-05,
+      "loss": 0.8741,
+      "step": 194220
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524456563133121e-05,
+      "loss": 0.8295,
+      "step": 194230
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524380398190321e-05,
+      "loss": 0.8396,
+      "step": 194240
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5243042332475214e-05,
+      "loss": 0.9054,
+      "step": 194250
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5242280683047204e-05,
+      "loss": 0.8878,
+      "step": 194260
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.524151903361921e-05,
+      "loss": 0.9178,
+      "step": 194270
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5240757384191206e-05,
+      "loss": 1.0165,
+      "step": 194280
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5239995734763203e-05,
+      "loss": 0.8226,
+      "step": 194290
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.52392340853352e-05,
+      "loss": 0.924,
+      "step": 194300
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5238472435907205e-05,
+      "loss": 0.7544,
+      "step": 194310
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.52377107864792e-05,
+      "loss": 0.8379,
+      "step": 194320
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.52369491370512e-05,
+      "loss": 1.035,
+      "step": 194330
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5236187487623204e-05,
+      "loss": 0.8191,
+      "step": 194340
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.5235425838195195e-05,
+      "loss": 0.9276,
+      "step": 194350
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.52346641887672e-05,
+      "loss": 0.9234,
+      "step": 194360
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.523390253933919e-05,
+      "loss": 0.8954,
+      "step": 194370
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.5233140889911194e-05,
+      "loss": 0.9798,
+      "step": 194380
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.523237924048319e-05,
+      "loss": 0.9574,
+      "step": 194390
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.523161759105519e-05,
+      "loss": 0.8137,
+      "step": 194400
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.523085594162719e-05,
+      "loss": 0.8339,
+      "step": 194410
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.523009429219919e-05,
+      "loss": 0.948,
+      "step": 194420
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522933264277118e-05,
+      "loss": 0.8428,
+      "step": 194430
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.5228570993343186e-05,
+      "loss": 0.8693,
+      "step": 194440
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522780934391518e-05,
+      "loss": 0.9482,
+      "step": 194450
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522704769448718e-05,
+      "loss": 0.8522,
+      "step": 194460
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522628604505918e-05,
+      "loss": 0.81,
+      "step": 194470
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522552439563118e-05,
+      "loss": 0.8471,
+      "step": 194480
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522476274620318e-05,
+      "loss": 0.7505,
+      "step": 194490
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522400109677518e-05,
+      "loss": 1.011,
+      "step": 194500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522323944734718e-05,
+      "loss": 0.8639,
+      "step": 194510
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522247779791917e-05,
+      "loss": 0.8816,
+      "step": 194520
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.5221716148491177e-05,
+      "loss": 0.8558,
+      "step": 194530
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.5220954499063174e-05,
+      "loss": 0.8996,
+      "step": 194540
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.522019284963517e-05,
+      "loss": 0.9194,
+      "step": 194550
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.521943120020717e-05,
+      "loss": 0.8934,
+      "step": 194560
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.521866955077917e-05,
+      "loss": 0.8327,
+      "step": 194570
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5217907901351164e-05,
+      "loss": 0.9396,
+      "step": 194580
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.521714625192317e-05,
+      "loss": 0.8424,
+      "step": 194590
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5216384602495166e-05,
+      "loss": 0.8828,
+      "step": 194600
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.521562295306716e-05,
+      "loss": 0.9673,
+      "step": 194610
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.521486130363916e-05,
+      "loss": 0.8517,
+      "step": 194620
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5214099654211165e-05,
+      "loss": 0.8744,
+      "step": 194630
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5213338004783155e-05,
+      "loss": 0.9084,
+      "step": 194640
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.521257635535516e-05,
+      "loss": 0.9371,
+      "step": 194650
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.521181470592716e-05,
+      "loss": 0.9466,
+      "step": 194660
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5211053056499155e-05,
+      "loss": 0.9296,
+      "step": 194670
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.521029140707115e-05,
+      "loss": 0.7993,
+      "step": 194680
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5209529757643156e-05,
+      "loss": 0.945,
+      "step": 194690
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5208768108215154e-05,
+      "loss": 0.8125,
+      "step": 194700
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.520800645878715e-05,
+      "loss": 0.8245,
+      "step": 194710
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5207244809359156e-05,
+      "loss": 0.7995,
+      "step": 194720
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5206483159931146e-05,
+      "loss": 0.8821,
+      "step": 194730
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.520572151050315e-05,
+      "loss": 0.8436,
+      "step": 194740
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.520495986107515e-05,
+      "loss": 0.8834,
+      "step": 194750
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.5204198211647145e-05,
+      "loss": 0.9409,
+      "step": 194760
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.520343656221914e-05,
+      "loss": 0.84,
+      "step": 194770
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.520267491279115e-05,
+      "loss": 0.9095,
+      "step": 194780
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.520191326336314e-05,
+      "loss": 0.8928,
+      "step": 194790
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.520115161393514e-05,
+      "loss": 0.9496,
+      "step": 194800
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.520038996450714e-05,
+      "loss": 0.8034,
+      "step": 194810
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519962831507914e-05,
+      "loss": 0.9712,
+      "step": 194820
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5198866665651134e-05,
+      "loss": 0.8936,
+      "step": 194830
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519810501622313e-05,
+      "loss": 0.8308,
+      "step": 194840
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519734336679513e-05,
+      "loss": 0.8723,
+      "step": 194850
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5196581717367134e-05,
+      "loss": 0.9363,
+      "step": 194860
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519582006793913e-05,
+      "loss": 0.8597,
+      "step": 194870
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519505841851113e-05,
+      "loss": 0.9182,
+      "step": 194880
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5194296769083126e-05,
+      "loss": 0.8831,
+      "step": 194890
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5193535119655124e-05,
+      "loss": 0.9009,
+      "step": 194900
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519277347022713e-05,
+      "loss": 0.8589,
+      "step": 194910
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5192011820799125e-05,
+      "loss": 0.791,
+      "step": 194920
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519125017137112e-05,
+      "loss": 0.7897,
+      "step": 194930
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.519048852194312e-05,
+      "loss": 0.8496,
+      "step": 194940
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5189726872515124e-05,
+      "loss": 0.847,
+      "step": 194950
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5188965223087115e-05,
+      "loss": 0.9106,
+      "step": 194960
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.518820357365912e-05,
+      "loss": 0.9449,
+      "step": 194970
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.518744192423112e-05,
+      "loss": 0.9381,
+      "step": 194980
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5186680274803114e-05,
+      "loss": 0.8421,
+      "step": 194990
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.518591862537511e-05,
+      "loss": 0.8855,
+      "step": 195000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5185156975947116e-05,
+      "loss": 0.8802,
+      "step": 195010
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.518439532651911e-05,
+      "loss": 0.9021,
+      "step": 195020
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.518363367709111e-05,
+      "loss": 0.8804,
+      "step": 195030
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.518287202766311e-05,
+      "loss": 0.9171,
+      "step": 195040
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5182110378235106e-05,
+      "loss": 0.994,
+      "step": 195050
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51813487288071e-05,
+      "loss": 0.8427,
+      "step": 195060
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.518058707937911e-05,
+      "loss": 0.8852,
+      "step": 195070
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5179825429951105e-05,
+      "loss": 0.8724,
+      "step": 195080
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51790637805231e-05,
+      "loss": 0.8652,
+      "step": 195090
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51783021310951e-05,
+      "loss": 0.9091,
+      "step": 195100
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51775404816671e-05,
+      "loss": 0.9727,
+      "step": 195110
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51767788322391e-05,
+      "loss": 0.9037,
+      "step": 195120
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51760171828111e-05,
+      "loss": 0.8059,
+      "step": 195130
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5175255533383097e-05,
+      "loss": 0.865,
+      "step": 195140
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5174493883955094e-05,
+      "loss": 0.9339,
+      "step": 195150
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.51737322345271e-05,
+      "loss": 0.9122,
+      "step": 195160
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.517297058509909e-05,
+      "loss": 1.0008,
+      "step": 195170
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.517220893567109e-05,
+      "loss": 0.8479,
+      "step": 195180
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.517144728624309e-05,
+      "loss": 1.0024,
+      "step": 195190
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.517068563681509e-05,
+      "loss": 0.923,
+      "step": 195200
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5169923987387086e-05,
+      "loss": 0.8796,
+      "step": 195210
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.516916233795909e-05,
+      "loss": 0.9672,
+      "step": 195220
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.516840068853108e-05,
+      "loss": 0.9133,
+      "step": 195230
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5167639039103085e-05,
+      "loss": 0.8725,
+      "step": 195240
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.516687738967508e-05,
+      "loss": 0.9532,
+      "step": 195250
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.516611574024708e-05,
+      "loss": 0.8182,
+      "step": 195260
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.516535409081908e-05,
+      "loss": 0.8759,
+      "step": 195270
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5164592441391075e-05,
+      "loss": 0.9639,
+      "step": 195280
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.516383079196308e-05,
+      "loss": 0.9514,
+      "step": 195290
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5163069142535076e-05,
+      "loss": 0.9031,
+      "step": 195300
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5162307493107074e-05,
+      "loss": 0.7907,
+      "step": 195310
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.516154584367907e-05,
+      "loss": 0.8834,
+      "step": 195320
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5160784194251076e-05,
+      "loss": 0.8051,
+      "step": 195330
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5160022544823066e-05,
+      "loss": 0.8541,
+      "step": 195340
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515926089539507e-05,
+      "loss": 0.8656,
+      "step": 195350
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515849924596707e-05,
+      "loss": 0.9527,
+      "step": 195360
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5157737596539065e-05,
+      "loss": 0.8719,
+      "step": 195370
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515697594711106e-05,
+      "loss": 0.8991,
+      "step": 195380
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515621429768307e-05,
+      "loss": 0.8472,
+      "step": 195390
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515545264825506e-05,
+      "loss": 0.9608,
+      "step": 195400
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515469099882706e-05,
+      "loss": 0.9535,
+      "step": 195410
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515392934939906e-05,
+      "loss": 0.8663,
+      "step": 195420
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515316769997106e-05,
+      "loss": 0.9931,
+      "step": 195430
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5152406050543054e-05,
+      "loss": 0.878,
+      "step": 195440
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515164440111506e-05,
+      "loss": 0.8786,
+      "step": 195450
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5150882751687056e-05,
+      "loss": 0.9401,
+      "step": 195460
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5150121102259054e-05,
+      "loss": 0.7812,
+      "step": 195470
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514935945283105e-05,
+      "loss": 0.9409,
+      "step": 195480
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514859780340305e-05,
+      "loss": 0.8859,
+      "step": 195490
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514783615397505e-05,
+      "loss": 0.8784,
+      "step": 195500
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514707450454705e-05,
+      "loss": 0.8177,
+      "step": 195510
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514631285511905e-05,
+      "loss": 0.9658,
+      "step": 195520
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5145551205691045e-05,
+      "loss": 0.9271,
+      "step": 195530
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514478955626305e-05,
+      "loss": 0.9874,
+      "step": 195540
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514402790683504e-05,
+      "loss": 0.9226,
+      "step": 195550
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5143266257407044e-05,
+      "loss": 0.8886,
+      "step": 195560
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514250460797904e-05,
+      "loss": 0.9035,
+      "step": 195570
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514174295855104e-05,
+      "loss": 0.9038,
+      "step": 195580
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514098130912304e-05,
+      "loss": 0.9301,
+      "step": 195590
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.514021965969504e-05,
+      "loss": 0.9641,
+      "step": 195600
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.513945801026703e-05,
+      "loss": 0.8509,
+      "step": 195610
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5138696360839036e-05,
+      "loss": 0.794,
+      "step": 195620
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5137934711411033e-05,
+      "loss": 0.8713,
+      "step": 195630
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.513717306198303e-05,
+      "loss": 0.9363,
+      "step": 195640
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.513641141255503e-05,
+      "loss": 0.8966,
+      "step": 195650
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.513564976312703e-05,
+      "loss": 1.0117,
+      "step": 195660
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.513488811369903e-05,
+      "loss": 0.8648,
+      "step": 195670
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.513412646427103e-05,
+      "loss": 0.9722,
+      "step": 195680
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5133364814843025e-05,
+      "loss": 1.0097,
+      "step": 195690
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.513260316541502e-05,
+      "loss": 0.9101,
+      "step": 195700
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.513184151598703e-05,
+      "loss": 0.8037,
+      "step": 195710
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5131079866559024e-05,
+      "loss": 0.8536,
+      "step": 195720
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.513031821713102e-05,
+      "loss": 0.8978,
+      "step": 195730
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512955656770302e-05,
+      "loss": 0.8586,
+      "step": 195740
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5128794918275017e-05,
+      "loss": 0.8439,
+      "step": 195750
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5128033268847014e-05,
+      "loss": 0.8414,
+      "step": 195760
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512727161941902e-05,
+      "loss": 0.8804,
+      "step": 195770
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512650996999101e-05,
+      "loss": 0.9829,
+      "step": 195780
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512574832056301e-05,
+      "loss": 0.9582,
+      "step": 195790
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512498667113501e-05,
+      "loss": 0.9216,
+      "step": 195800
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512422502170701e-05,
+      "loss": 1.0311,
+      "step": 195810
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5123463372279006e-05,
+      "loss": 0.8859,
+      "step": 195820
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512270172285101e-05,
+      "loss": 0.9172,
+      "step": 195830
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.512194007342301e-05,
+      "loss": 0.8642,
+      "step": 195840
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5121178423995005e-05,
+      "loss": 0.9191,
+      "step": 195850
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5120416774567e-05,
+      "loss": 0.9679,
+      "step": 195860
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5119655125139e-05,
+      "loss": 0.852,
+      "step": 195870
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5118893475711004e-05,
+      "loss": 0.8749,
+      "step": 195880
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5118131826283e-05,
+      "loss": 0.9241,
+      "step": 195890
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5117370176855e-05,
+      "loss": 0.8602,
+      "step": 195900
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5116608527426996e-05,
+      "loss": 0.8374,
+      "step": 195910
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5115846877999e-05,
+      "loss": 0.8507,
+      "step": 195920
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.511508522857099e-05,
+      "loss": 0.9574,
+      "step": 195930
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5114323579142996e-05,
+      "loss": 0.9202,
+      "step": 195940
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.511356192971499e-05,
+      "loss": 0.9301,
+      "step": 195950
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.511280028028699e-05,
+      "loss": 0.9926,
+      "step": 195960
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.511203863085899e-05,
+      "loss": 0.8906,
+      "step": 195970
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.511127698143099e-05,
+      "loss": 0.8729,
+      "step": 195980
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.511051533200298e-05,
+      "loss": 0.9529,
+      "step": 195990
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510975368257499e-05,
+      "loss": 0.892,
+      "step": 196000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5108992033146985e-05,
+      "loss": 0.8837,
+      "step": 196010
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510823038371898e-05,
+      "loss": 0.8873,
+      "step": 196020
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510746873429098e-05,
+      "loss": 1.0603,
+      "step": 196030
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5106707084862984e-05,
+      "loss": 0.8367,
+      "step": 196040
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510594543543498e-05,
+      "loss": 0.8823,
+      "step": 196050
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510518378600698e-05,
+      "loss": 0.8182,
+      "step": 196060
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5104422136578976e-05,
+      "loss": 0.9323,
+      "step": 196070
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5103660487150974e-05,
+      "loss": 0.9818,
+      "step": 196080
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510289883772298e-05,
+      "loss": 0.9048,
+      "step": 196090
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5102137188294975e-05,
+      "loss": 0.868,
+      "step": 196100
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.510137553886697e-05,
+      "loss": 0.9462,
+      "step": 196110
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.510061388943897e-05,
+      "loss": 0.8624,
+      "step": 196120
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5099852240010975e-05,
+      "loss": 0.9142,
+      "step": 196130
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5099090590582965e-05,
+      "loss": 0.9252,
+      "step": 196140
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509832894115497e-05,
+      "loss": 0.8638,
+      "step": 196150
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509756729172697e-05,
+      "loss": 0.9475,
+      "step": 196160
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5096805642298964e-05,
+      "loss": 0.9503,
+      "step": 196170
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509604399287096e-05,
+      "loss": 0.8782,
+      "step": 196180
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5095282343442966e-05,
+      "loss": 0.8909,
+      "step": 196190
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509452069401496e-05,
+      "loss": 0.9343,
+      "step": 196200
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509375904458696e-05,
+      "loss": 0.9718,
+      "step": 196210
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509299739515896e-05,
+      "loss": 0.8825,
+      "step": 196220
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5092235745730956e-05,
+      "loss": 0.8348,
+      "step": 196230
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5091474096302953e-05,
+      "loss": 0.8415,
+      "step": 196240
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.509071244687495e-05,
+      "loss": 0.9613,
+      "step": 196250
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5089950797446955e-05,
+      "loss": 0.8555,
+      "step": 196260
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.508918914801895e-05,
+      "loss": 0.9311,
+      "step": 196270
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.508842749859095e-05,
+      "loss": 0.8881,
+      "step": 196280
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.508766584916295e-05,
+      "loss": 0.9557,
+      "step": 196290
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.508690419973495e-05,
+      "loss": 0.8168,
+      "step": 196300
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.508614255030694e-05,
+      "loss": 0.8917,
+      "step": 196310
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.508538090087895e-05,
+      "loss": 0.8555,
+      "step": 196320
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.5084619251450944e-05,
+      "loss": 0.955,
+      "step": 196330
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.508385760202294e-05,
+      "loss": 0.8221,
+      "step": 196340
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.508309595259494e-05,
+      "loss": 0.9489,
+      "step": 196350
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.508233430316694e-05,
+      "loss": 0.9746,
+      "step": 196360
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5081572653738934e-05,
+      "loss": 0.899,
+      "step": 196370
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.508081100431094e-05,
+      "loss": 0.824,
+      "step": 196380
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5080049354882936e-05,
+      "loss": 0.8202,
+      "step": 196390
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507928770545493e-05,
+      "loss": 0.9546,
+      "step": 196400
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507852605602693e-05,
+      "loss": 0.8689,
+      "step": 196410
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5077764406598935e-05,
+      "loss": 0.8061,
+      "step": 196420
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507700275717093e-05,
+      "loss": 0.8143,
+      "step": 196430
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507624110774293e-05,
+      "loss": 0.961,
+      "step": 196440
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507547945831493e-05,
+      "loss": 0.9381,
+      "step": 196450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5074717808886925e-05,
+      "loss": 0.9024,
+      "step": 196460
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507395615945893e-05,
+      "loss": 0.9912,
+      "step": 196470
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5073194510030927e-05,
+      "loss": 0.8402,
+      "step": 196480
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5072432860602924e-05,
+      "loss": 0.9538,
+      "step": 196490
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.507167121117492e-05,
+      "loss": 0.8891,
+      "step": 196500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5070909561746926e-05,
+      "loss": 0.915,
+      "step": 196510
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5070147912318916e-05,
+      "loss": 0.8352,
+      "step": 196520
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.506938626289092e-05,
+      "loss": 0.9149,
+      "step": 196530
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.506862461346292e-05,
+      "loss": 0.9036,
+      "step": 196540
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.5067862964034916e-05,
+      "loss": 0.8475,
+      "step": 196550
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506710131460691e-05,
+      "loss": 0.8192,
+      "step": 196560
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506633966517892e-05,
+      "loss": 0.9789,
+      "step": 196570
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506557801575091e-05,
+      "loss": 0.9161,
+      "step": 196580
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506481636632291e-05,
+      "loss": 0.9072,
+      "step": 196590
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506405471689491e-05,
+      "loss": 1.0558,
+      "step": 196600
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506329306746691e-05,
+      "loss": 0.8995,
+      "step": 196610
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5062531418038905e-05,
+      "loss": 0.8484,
+      "step": 196620
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.506176976861091e-05,
+      "loss": 0.8247,
+      "step": 196630
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5061008119182906e-05,
+      "loss": 0.9305,
+      "step": 196640
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5060246469754904e-05,
+      "loss": 0.9195,
+      "step": 196650
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.50594848203269e-05,
+      "loss": 0.9872,
+      "step": 196660
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.50587231708989e-05,
+      "loss": 1.0031,
+      "step": 196670
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.50579615214709e-05,
+      "loss": 0.8999,
+      "step": 196680
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5057199872042894e-05,
+      "loss": 1.0099,
+      "step": 196690
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.50564382226149e-05,
+      "loss": 0.8442,
+      "step": 196700
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5055676573186895e-05,
+      "loss": 0.8904,
+      "step": 196710
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.505491492375889e-05,
+      "loss": 0.8055,
+      "step": 196720
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.505415327433089e-05,
+      "loss": 0.7827,
+      "step": 196730
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5053391624902895e-05,
+      "loss": 0.8712,
+      "step": 196740
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.5052629975474885e-05,
+      "loss": 0.8839,
+      "step": 196750
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.505186832604689e-05,
+      "loss": 0.8258,
+      "step": 196760
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.505110667661889e-05,
+      "loss": 1.0771,
+      "step": 196770
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5050345027190884e-05,
+      "loss": 0.8884,
+      "step": 196780
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504958337776288e-05,
+      "loss": 0.7982,
+      "step": 196790
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5048821728334886e-05,
+      "loss": 0.917,
+      "step": 196800
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5048060078906884e-05,
+      "loss": 0.9178,
+      "step": 196810
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504729842947888e-05,
+      "loss": 0.9027,
+      "step": 196820
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504653678005088e-05,
+      "loss": 0.8165,
+      "step": 196830
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5045775130622876e-05,
+      "loss": 0.9814,
+      "step": 196840
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504501348119488e-05,
+      "loss": 0.8673,
+      "step": 196850
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504425183176688e-05,
+      "loss": 0.9013,
+      "step": 196860
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5043490182338875e-05,
+      "loss": 0.9198,
+      "step": 196870
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504272853291087e-05,
+      "loss": 0.8511,
+      "step": 196880
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504196688348288e-05,
+      "loss": 0.8538,
+      "step": 196890
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504120523405487e-05,
+      "loss": 0.8871,
+      "step": 196900
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.504044358462687e-05,
+      "loss": 0.9141,
+      "step": 196910
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.503968193519887e-05,
+      "loss": 0.8776,
+      "step": 196920
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.503892028577087e-05,
+      "loss": 0.9449,
+      "step": 196930
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5038158636342864e-05,
+      "loss": 1.0362,
+      "step": 196940
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.503739698691487e-05,
+      "loss": 0.8496,
+      "step": 196950
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.503663533748686e-05,
+      "loss": 0.8386,
+      "step": 196960
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.5035873688058863e-05,
+      "loss": 0.8338,
+      "step": 196970
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.503511203863086e-05,
+      "loss": 0.885,
+      "step": 196980
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.503435038920286e-05,
+      "loss": 0.8631,
+      "step": 196990
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5033588739774856e-05,
+      "loss": 0.9078,
+      "step": 197000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.503282709034686e-05,
+      "loss": 0.7839,
+      "step": 197010
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.503206544091886e-05,
+      "loss": 0.8536,
+      "step": 197020
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5031303791490855e-05,
+      "loss": 0.8495,
+      "step": 197030
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.503054214206285e-05,
+      "loss": 0.8198,
+      "step": 197040
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502978049263485e-05,
+      "loss": 0.9468,
+      "step": 197050
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5029018843206854e-05,
+      "loss": 0.8783,
+      "step": 197060
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502825719377885e-05,
+      "loss": 1.0395,
+      "step": 197070
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502749554435085e-05,
+      "loss": 0.8007,
+      "step": 197080
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5026733894922847e-05,
+      "loss": 0.9032,
+      "step": 197090
+    },
+    {
+      "epoch": 9.0,
+      "eval_cer": 0.6203580231947307,
+      "eval_em": 0.007568590350047304,
+      "eval_f1": 0.007568590350047304,
+      "eval_loss": 0.8918395638465881,
+      "eval_runtime": 1011.3782,
+      "eval_samples_per_second": 10.451,
+      "eval_steps_per_second": 1.307,
+      "eval_wer": 0.9924314096499527,
+      "step": 197097
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502597224549485e-05,
+      "loss": 0.9247,
+      "step": 197100
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502521059606684e-05,
+      "loss": 0.8504,
+      "step": 197110
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5024448946638846e-05,
+      "loss": 0.8581,
+      "step": 197120
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5023687297210836e-05,
+      "loss": 0.8436,
+      "step": 197130
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502292564778284e-05,
+      "loss": 0.8148,
+      "step": 197140
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502216399835484e-05,
+      "loss": 0.8869,
+      "step": 197150
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.5021402348926836e-05,
+      "loss": 0.8395,
+      "step": 197160
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.502064069949883e-05,
+      "loss": 0.8322,
+      "step": 197170
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.501987905007084e-05,
+      "loss": 0.9458,
+      "step": 197180
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.501911740064283e-05,
+      "loss": 0.7932,
+      "step": 197190
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.501835575121483e-05,
+      "loss": 0.8055,
+      "step": 197200
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501759410178683e-05,
+      "loss": 0.9325,
+      "step": 197210
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501683245235883e-05,
+      "loss": 0.7799,
+      "step": 197220
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501607080293083e-05,
+      "loss": 0.7827,
+      "step": 197230
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501530915350283e-05,
+      "loss": 0.829,
+      "step": 197240
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.5014547504074826e-05,
+      "loss": 0.8383,
+      "step": 197250
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.5013785854646824e-05,
+      "loss": 0.853,
+      "step": 197260
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501302420521883e-05,
+      "loss": 0.7857,
+      "step": 197270
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501226255579082e-05,
+      "loss": 0.9044,
+      "step": 197280
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501150090636282e-05,
+      "loss": 0.8442,
+      "step": 197290
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.501073925693482e-05,
+      "loss": 0.8541,
+      "step": 197300
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500997760750682e-05,
+      "loss": 0.8579,
+      "step": 197310
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.5009215958078815e-05,
+      "loss": 0.8604,
+      "step": 197320
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500845430865082e-05,
+      "loss": 0.7805,
+      "step": 197330
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500769265922281e-05,
+      "loss": 0.911,
+      "step": 197340
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.5006931009794815e-05,
+      "loss": 0.9231,
+      "step": 197350
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500616936036681e-05,
+      "loss": 0.8556,
+      "step": 197360
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500540771093881e-05,
+      "loss": 0.8509,
+      "step": 197370
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500464606151081e-05,
+      "loss": 0.8298,
+      "step": 197380
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.500388441208281e-05,
+      "loss": 0.8365,
+      "step": 197390
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.50031227626548e-05,
+      "loss": 0.9822,
+      "step": 197400
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.5002361113226806e-05,
+      "loss": 0.7865,
+      "step": 197410
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.5001599463798804e-05,
+      "loss": 0.7906,
+      "step": 197420
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.50008378143708e-05,
+      "loss": 0.7339,
+      "step": 197430
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.5000076164942805e-05,
+      "loss": 0.817,
+      "step": 197440
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.49993145155148e-05,
+      "loss": 0.8486,
+      "step": 197450
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.49985528660868e-05,
+      "loss": 0.8249,
+      "step": 197460
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.49977912166588e-05,
+      "loss": 0.8901,
+      "step": 197470
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.49970295672308e-05,
+      "loss": 0.8891,
+      "step": 197480
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.499626791780279e-05,
+      "loss": 0.8489,
+      "step": 197490
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.49955062683748e-05,
+      "loss": 0.8218,
+      "step": 197500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.4994744618946794e-05,
+      "loss": 0.8507,
+      "step": 197510
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.499398296951879e-05,
+      "loss": 0.817,
+      "step": 197520
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.499322132009079e-05,
+      "loss": 0.8764,
+      "step": 197530
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.4992459670662794e-05,
+      "loss": 0.9023,
+      "step": 197540
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.4991698021234784e-05,
+      "loss": 0.9322,
+      "step": 197550
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.499093637180679e-05,
+      "loss": 0.8837,
+      "step": 197560
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.4990174722378786e-05,
+      "loss": 0.773,
+      "step": 197570
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.4989413072950783e-05,
+      "loss": 0.8909,
+      "step": 197580
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.498865142352278e-05,
+      "loss": 0.8686,
+      "step": 197590
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.498788977409478e-05,
+      "loss": 0.8659,
+      "step": 197600
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.498712812466678e-05,
+      "loss": 0.9429,
+      "step": 197610
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.498636647523878e-05,
+      "loss": 0.8642,
+      "step": 197620
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.498560482581078e-05,
+      "loss": 0.8397,
+      "step": 197630
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.4984843176382775e-05,
+      "loss": 0.7975,
+      "step": 197640
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.498408152695478e-05,
+      "loss": 0.8546,
+      "step": 197650
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.498331987752677e-05,
+      "loss": 0.8227,
+      "step": 197660
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.4982558228098774e-05,
+      "loss": 0.8103,
+      "step": 197670
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.498179657867077e-05,
+      "loss": 0.8694,
+      "step": 197680
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.498103492924277e-05,
+      "loss": 0.8483,
+      "step": 197690
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.4980273279814767e-05,
+      "loss": 0.8387,
+      "step": 197700
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497951163038677e-05,
+      "loss": 0.8343,
+      "step": 197710
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497874998095876e-05,
+      "loss": 0.8556,
+      "step": 197720
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.4977988331530766e-05,
+      "loss": 0.8354,
+      "step": 197730
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497722668210276e-05,
+      "loss": 0.8617,
+      "step": 197740
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497646503267476e-05,
+      "loss": 0.8637,
+      "step": 197750
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497570338324676e-05,
+      "loss": 0.8961,
+      "step": 197760
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497494173381876e-05,
+      "loss": 0.8148,
+      "step": 197770
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497418008439075e-05,
+      "loss": 0.8909,
+      "step": 197780
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497341843496276e-05,
+      "loss": 0.8685,
+      "step": 197790
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.4972656785534755e-05,
+      "loss": 0.8247,
+      "step": 197800
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.497189513610675e-05,
+      "loss": 0.86,
+      "step": 197810
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.4971133486678756e-05,
+      "loss": 0.8419,
+      "step": 197820
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.4970371837250754e-05,
+      "loss": 0.8225,
+      "step": 197830
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.496961018782275e-05,
+      "loss": 0.9576,
+      "step": 197840
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.496884853839475e-05,
+      "loss": 0.8606,
+      "step": 197850
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.496808688896675e-05,
+      "loss": 0.9276,
+      "step": 197860
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4967325239538744e-05,
+      "loss": 0.9294,
+      "step": 197870
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.496656359011075e-05,
+      "loss": 0.9438,
+      "step": 197880
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4965801940682746e-05,
+      "loss": 0.8802,
+      "step": 197890
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.496504029125474e-05,
+      "loss": 0.9007,
+      "step": 197900
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.496427864182674e-05,
+      "loss": 0.8481,
+      "step": 197910
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4963516992398745e-05,
+      "loss": 0.9469,
+      "step": 197920
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4962755342970735e-05,
+      "loss": 0.7888,
+      "step": 197930
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.496199369354274e-05,
+      "loss": 0.8167,
+      "step": 197940
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.496123204411474e-05,
+      "loss": 0.8602,
+      "step": 197950
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4960470394686735e-05,
+      "loss": 0.8369,
+      "step": 197960
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495970874525873e-05,
+      "loss": 0.793,
+      "step": 197970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4958947095830736e-05,
+      "loss": 0.9152,
+      "step": 197980
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495818544640273e-05,
+      "loss": 0.8685,
+      "step": 197990
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495742379697473e-05,
+      "loss": 0.8499,
+      "step": 198000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495666214754673e-05,
+      "loss": 0.7673,
+      "step": 198010
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4955900498118726e-05,
+      "loss": 0.8627,
+      "step": 198020
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495513884869073e-05,
+      "loss": 0.9134,
+      "step": 198030
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495437719926273e-05,
+      "loss": 0.8363,
+      "step": 198040
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.4953615549834725e-05,
+      "loss": 0.8884,
+      "step": 198050
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495285390040672e-05,
+      "loss": 0.8133,
+      "step": 198060
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495209225097872e-05,
+      "loss": 0.7654,
+      "step": 198070
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.495133060155072e-05,
+      "loss": 0.9106,
+      "step": 198080
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.495056895212272e-05,
+      "loss": 0.909,
+      "step": 198090
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.494980730269471e-05,
+      "loss": 0.9101,
+      "step": 198100
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.494904565326672e-05,
+      "loss": 0.9155,
+      "step": 198110
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4948284003838714e-05,
+      "loss": 0.7244,
+      "step": 198120
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.494752235441071e-05,
+      "loss": 0.8359,
+      "step": 198130
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.494676070498271e-05,
+      "loss": 0.8449,
+      "step": 198140
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4945999055554714e-05,
+      "loss": 1.0197,
+      "step": 198150
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4945237406126704e-05,
+      "loss": 0.8197,
+      "step": 198160
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.494447575669871e-05,
+      "loss": 0.9156,
+      "step": 198170
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4943714107270706e-05,
+      "loss": 0.929,
+      "step": 198180
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4942952457842703e-05,
+      "loss": 0.8944,
+      "step": 198190
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.49421908084147e-05,
+      "loss": 0.8605,
+      "step": 198200
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4941429158986705e-05,
+      "loss": 0.8782,
+      "step": 198210
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.49406675095587e-05,
+      "loss": 0.8471,
+      "step": 198220
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.49399058601307e-05,
+      "loss": 0.9093,
+      "step": 198230
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4939144210702704e-05,
+      "loss": 0.7729,
+      "step": 198240
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4938382561274695e-05,
+      "loss": 0.853,
+      "step": 198250
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.49376209118467e-05,
+      "loss": 0.8386,
+      "step": 198260
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.49368592624187e-05,
+      "loss": 0.8352,
+      "step": 198270
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4936097612990694e-05,
+      "loss": 0.89,
+      "step": 198280
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.493533596356269e-05,
+      "loss": 0.848,
+      "step": 198290
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.4934574314134696e-05,
+      "loss": 0.9717,
+      "step": 198300
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.4933812664706687e-05,
+      "loss": 0.7322,
+      "step": 198310
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.493305101527869e-05,
+      "loss": 0.8249,
+      "step": 198320
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.493228936585069e-05,
+      "loss": 0.7863,
+      "step": 198330
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.4931527716422686e-05,
+      "loss": 0.7867,
+      "step": 198340
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.493076606699468e-05,
+      "loss": 0.8813,
+      "step": 198350
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.493000441756669e-05,
+      "loss": 0.8001,
+      "step": 198360
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492924276813868e-05,
+      "loss": 0.9422,
+      "step": 198370
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492848111871068e-05,
+      "loss": 0.8662,
+      "step": 198380
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492771946928268e-05,
+      "loss": 0.86,
+      "step": 198390
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492695781985468e-05,
+      "loss": 0.8411,
+      "step": 198400
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492619617042668e-05,
+      "loss": 0.8704,
+      "step": 198410
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492543452099868e-05,
+      "loss": 0.8177,
+      "step": 198420
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.4924672871570676e-05,
+      "loss": 0.9208,
+      "step": 198430
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.4923911222142674e-05,
+      "loss": 0.8759,
+      "step": 198440
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492314957271468e-05,
+      "loss": 0.8519,
+      "step": 198450
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492238792328667e-05,
+      "loss": 0.8614,
+      "step": 198460
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492162627385867e-05,
+      "loss": 0.8454,
+      "step": 198470
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492086462443067e-05,
+      "loss": 0.8127,
+      "step": 198480
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.492010297500267e-05,
+      "loss": 0.8054,
+      "step": 198490
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.4919341325574666e-05,
+      "loss": 0.8115,
+      "step": 198500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.491857967614666e-05,
+      "loss": 0.9805,
+      "step": 198510
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.491781802671866e-05,
+      "loss": 0.8246,
+      "step": 198520
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4917056377290665e-05,
+      "loss": 0.8857,
+      "step": 198530
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4916294727862655e-05,
+      "loss": 0.8634,
+      "step": 198540
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.491553307843466e-05,
+      "loss": 0.9155,
+      "step": 198550
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.491477142900666e-05,
+      "loss": 0.7713,
+      "step": 198560
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4914009779578655e-05,
+      "loss": 0.8689,
+      "step": 198570
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.491324813015065e-05,
+      "loss": 0.8533,
+      "step": 198580
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4912486480722656e-05,
+      "loss": 0.917,
+      "step": 198590
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4911724831294654e-05,
+      "loss": 0.8913,
+      "step": 198600
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.491096318186665e-05,
+      "loss": 0.8416,
+      "step": 198610
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4910201532438655e-05,
+      "loss": 0.8362,
+      "step": 198620
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4909439883010646e-05,
+      "loss": 0.7731,
+      "step": 198630
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490867823358265e-05,
+      "loss": 1.0091,
+      "step": 198640
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490791658415465e-05,
+      "loss": 0.8446,
+      "step": 198650
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4907154934726645e-05,
+      "loss": 0.8279,
+      "step": 198660
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490639328529864e-05,
+      "loss": 0.9431,
+      "step": 198670
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490563163587065e-05,
+      "loss": 0.9201,
+      "step": 198680
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490486998644264e-05,
+      "loss": 0.8349,
+      "step": 198690
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490410833701464e-05,
+      "loss": 0.8873,
+      "step": 198700
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490334668758664e-05,
+      "loss": 0.9892,
+      "step": 198710
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490258503815864e-05,
+      "loss": 0.8478,
+      "step": 198720
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.4901823388730634e-05,
+      "loss": 1.0619,
+      "step": 198730
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.490106173930264e-05,
+      "loss": 0.8093,
+      "step": 198740
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.490030008987463e-05,
+      "loss": 0.8865,
+      "step": 198750
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.4899538440446634e-05,
+      "loss": 0.8094,
+      "step": 198760
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489877679101863e-05,
+      "loss": 0.7933,
+      "step": 198770
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489801514159063e-05,
+      "loss": 0.8299,
+      "step": 198780
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.4897253492162626e-05,
+      "loss": 0.9216,
+      "step": 198790
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489649184273463e-05,
+      "loss": 0.9382,
+      "step": 198800
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489573019330663e-05,
+      "loss": 0.8196,
+      "step": 198810
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.4894968543878625e-05,
+      "loss": 0.8931,
+      "step": 198820
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489420689445063e-05,
+      "loss": 0.7466,
+      "step": 198830
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489344524502262e-05,
+      "loss": 0.8823,
+      "step": 198840
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.4892683595594624e-05,
+      "loss": 0.7342,
+      "step": 198850
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489192194616662e-05,
+      "loss": 0.9113,
+      "step": 198860
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489116029673862e-05,
+      "loss": 0.8756,
+      "step": 198870
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.489039864731062e-05,
+      "loss": 0.8442,
+      "step": 198880
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.488963699788262e-05,
+      "loss": 0.763,
+      "step": 198890
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.488887534845461e-05,
+      "loss": 0.7887,
+      "step": 198900
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.4888113699026616e-05,
+      "loss": 0.8963,
+      "step": 198910
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.488735204959861e-05,
+      "loss": 0.8016,
+      "step": 198920
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.488659040017061e-05,
+      "loss": 0.8261,
+      "step": 198930
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.488582875074261e-05,
+      "loss": 0.8482,
+      "step": 198940
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.488506710131461e-05,
+      "loss": 0.8855,
+      "step": 198950
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.48843054518866e-05,
+      "loss": 0.9307,
+      "step": 198960
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.488354380245861e-05,
+      "loss": 0.7758,
+      "step": 198970
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.4882782153030605e-05,
+      "loss": 0.9243,
+      "step": 198980
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.48820205036026e-05,
+      "loss": 0.9089,
+      "step": 198990
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.488125885417461e-05,
+      "loss": 0.8743,
+      "step": 199000
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.48804972047466e-05,
+      "loss": 0.8672,
+      "step": 199010
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.48797355553186e-05,
+      "loss": 0.8672,
+      "step": 199020
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.48789739058906e-05,
+      "loss": 0.8375,
+      "step": 199030
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.4878212256462597e-05,
+      "loss": 0.9095,
+      "step": 199040
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.4877450607034594e-05,
+      "loss": 0.8765,
+      "step": 199050
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.48766889576066e-05,
+      "loss": 0.7851,
+      "step": 199060
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.487592730817859e-05,
+      "loss": 0.8339,
+      "step": 199070
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.487516565875059e-05,
+      "loss": 0.94,
+      "step": 199080
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.487440400932259e-05,
+      "loss": 0.9685,
+      "step": 199090
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.487364235989459e-05,
+      "loss": 0.8154,
+      "step": 199100
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.4872880710466586e-05,
+      "loss": 0.8155,
+      "step": 199110
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.487211906103859e-05,
+      "loss": 0.8543,
+      "step": 199120
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.487135741161058e-05,
+      "loss": 0.8583,
+      "step": 199130
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.4870595762182585e-05,
+      "loss": 0.8356,
+      "step": 199140
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.486983411275458e-05,
+      "loss": 0.8735,
+      "step": 199150
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.486907246332658e-05,
+      "loss": 0.9047,
+      "step": 199160
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.486831081389858e-05,
+      "loss": 1.0363,
+      "step": 199170
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486754916447058e-05,
+      "loss": 0.8749,
+      "step": 199180
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486678751504258e-05,
+      "loss": 0.8864,
+      "step": 199190
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4866025865614576e-05,
+      "loss": 0.848,
+      "step": 199200
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486526421618658e-05,
+      "loss": 0.8558,
+      "step": 199210
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486450256675857e-05,
+      "loss": 0.8447,
+      "step": 199220
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4863740917330575e-05,
+      "loss": 0.949,
+      "step": 199230
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486297926790257e-05,
+      "loss": 0.8509,
+      "step": 199240
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486221761847457e-05,
+      "loss": 0.8239,
+      "step": 199250
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486145596904657e-05,
+      "loss": 0.9349,
+      "step": 199260
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.486069431961857e-05,
+      "loss": 0.8265,
+      "step": 199270
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.485993267019056e-05,
+      "loss": 0.9186,
+      "step": 199280
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.485917102076257e-05,
+      "loss": 0.9238,
+      "step": 199290
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4858409371334565e-05,
+      "loss": 0.962,
+      "step": 199300
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.485764772190656e-05,
+      "loss": 0.8981,
+      "step": 199310
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.485688607247856e-05,
+      "loss": 0.7876,
+      "step": 199320
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4856124423050564e-05,
+      "loss": 0.8965,
+      "step": 199330
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4855362773622554e-05,
+      "loss": 0.8428,
+      "step": 199340
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.485460112419456e-05,
+      "loss": 0.9359,
+      "step": 199350
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4853839474766556e-05,
+      "loss": 0.8981,
+      "step": 199360
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4853077825338554e-05,
+      "loss": 0.7856,
+      "step": 199370
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.485231617591055e-05,
+      "loss": 0.9532,
+      "step": 199380
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.4851554526482555e-05,
+      "loss": 0.8895,
+      "step": 199390
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.485079287705455e-05,
+      "loss": 0.7923,
+      "step": 199400
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.485003122762655e-05,
+      "loss": 0.9831,
+      "step": 199410
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.4849269578198554e-05,
+      "loss": 0.8158,
+      "step": 199420
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.4848507928770545e-05,
+      "loss": 0.8903,
+      "step": 199430
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484774627934255e-05,
+      "loss": 0.959,
+      "step": 199440
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484698462991454e-05,
+      "loss": 0.8866,
+      "step": 199450
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.4846222980486544e-05,
+      "loss": 0.8448,
+      "step": 199460
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484546133105854e-05,
+      "loss": 0.7669,
+      "step": 199470
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484469968163054e-05,
+      "loss": 0.9543,
+      "step": 199480
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484393803220254e-05,
+      "loss": 0.9012,
+      "step": 199490
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484317638277454e-05,
+      "loss": 0.8865,
+      "step": 199500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484241473334653e-05,
+      "loss": 0.9193,
+      "step": 199510
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.4841653083918536e-05,
+      "loss": 0.9453,
+      "step": 199520
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484089143449053e-05,
+      "loss": 0.8375,
+      "step": 199530
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.484012978506253e-05,
+      "loss": 0.8611,
+      "step": 199540
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.483936813563453e-05,
+      "loss": 0.9586,
+      "step": 199550
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.483860648620653e-05,
+      "loss": 0.8492,
+      "step": 199560
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.483784483677853e-05,
+      "loss": 0.8699,
+      "step": 199570
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.483708318735053e-05,
+      "loss": 0.9491,
+      "step": 199580
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.4836321537922525e-05,
+      "loss": 0.9336,
+      "step": 199590
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.483555988849452e-05,
+      "loss": 0.7788,
+      "step": 199600
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.483479823906653e-05,
+      "loss": 0.7707,
+      "step": 199610
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4834036589638524e-05,
+      "loss": 0.8131,
+      "step": 199620
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.483327494021052e-05,
+      "loss": 0.8256,
+      "step": 199630
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.483251329078252e-05,
+      "loss": 0.8455,
+      "step": 199640
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.483175164135452e-05,
+      "loss": 0.8345,
+      "step": 199650
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4830989991926514e-05,
+      "loss": 0.8885,
+      "step": 199660
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.483022834249852e-05,
+      "loss": 0.8918,
+      "step": 199670
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4829466693070516e-05,
+      "loss": 0.9028,
+      "step": 199680
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.482870504364251e-05,
+      "loss": 0.8143,
+      "step": 199690
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.482794339421451e-05,
+      "loss": 0.8265,
+      "step": 199700
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4827181744786515e-05,
+      "loss": 0.7252,
+      "step": 199710
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4826420095358506e-05,
+      "loss": 0.8068,
+      "step": 199720
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.482565844593051e-05,
+      "loss": 0.8609,
+      "step": 199730
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.482489679650251e-05,
+      "loss": 0.8168,
+      "step": 199740
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4824135147074505e-05,
+      "loss": 0.9892,
+      "step": 199750
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.48233734976465e-05,
+      "loss": 0.8258,
+      "step": 199760
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4822611848218506e-05,
+      "loss": 0.9542,
+      "step": 199770
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4821850198790504e-05,
+      "loss": 0.9344,
+      "step": 199780
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.48210885493625e-05,
+      "loss": 0.9838,
+      "step": 199790
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4820326899934506e-05,
+      "loss": 0.8579,
+      "step": 199800
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.4819565250506496e-05,
+      "loss": 0.8282,
+      "step": 199810
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.48188036010785e-05,
+      "loss": 0.8157,
+      "step": 199820
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.48180419516505e-05,
+      "loss": 0.817,
+      "step": 199830
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4817280302222495e-05,
+      "loss": 0.8555,
+      "step": 199840
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481651865279449e-05,
+      "loss": 0.8899,
+      "step": 199850
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.48157570033665e-05,
+      "loss": 0.7611,
+      "step": 199860
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481499535393849e-05,
+      "loss": 0.9506,
+      "step": 199870
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481423370451049e-05,
+      "loss": 0.8024,
+      "step": 199880
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481347205508248e-05,
+      "loss": 0.8506,
+      "step": 199890
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481271040565449e-05,
+      "loss": 0.9245,
+      "step": 199900
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4811948756226485e-05,
+      "loss": 0.8933,
+      "step": 199910
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481118710679848e-05,
+      "loss": 0.8346,
+      "step": 199920
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.481042545737048e-05,
+      "loss": 0.8426,
+      "step": 199930
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4809663807942484e-05,
+      "loss": 0.9583,
+      "step": 199940
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.480890215851448e-05,
+      "loss": 0.8746,
+      "step": 199950
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.480814050908648e-05,
+      "loss": 0.8754,
+      "step": 199960
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4807378859658476e-05,
+      "loss": 0.8315,
+      "step": 199970
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4806617210230474e-05,
+      "loss": 0.8255,
+      "step": 199980
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.480585556080248e-05,
+      "loss": 0.9084,
+      "step": 199990
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4805093911374475e-05,
+      "loss": 0.9409,
+      "step": 200000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.480433226194647e-05,
+      "loss": 0.8725,
+      "step": 200010
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.480357061251847e-05,
+      "loss": 0.8184,
+      "step": 200020
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4802808963090474e-05,
+      "loss": 0.9008,
+      "step": 200030
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.4802047313662465e-05,
+      "loss": 0.8914,
+      "step": 200040
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.480128566423447e-05,
+      "loss": 0.8669,
+      "step": 200050
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.480052401480647e-05,
+      "loss": 0.8301,
+      "step": 200060
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.4799762365378464e-05,
+      "loss": 0.9116,
+      "step": 200070
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479900071595046e-05,
+      "loss": 0.815,
+      "step": 200080
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.4798239066522466e-05,
+      "loss": 0.9064,
+      "step": 200090
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479747741709446e-05,
+      "loss": 0.8455,
+      "step": 200100
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479671576766646e-05,
+      "loss": 0.9071,
+      "step": 200110
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479595411823846e-05,
+      "loss": 0.8565,
+      "step": 200120
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.4795192468810456e-05,
+      "loss": 0.9023,
+      "step": 200130
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479443081938245e-05,
+      "loss": 0.829,
+      "step": 200140
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479366916995446e-05,
+      "loss": 0.8039,
+      "step": 200150
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.4792907520526455e-05,
+      "loss": 0.9015,
+      "step": 200160
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479214587109845e-05,
+      "loss": 0.779,
+      "step": 200170
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479138422167045e-05,
+      "loss": 0.9384,
+      "step": 200180
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.479062257224245e-05,
+      "loss": 0.8687,
+      "step": 200190
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478986092281445e-05,
+      "loss": 0.8365,
+      "step": 200200
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478909927338645e-05,
+      "loss": 0.8779,
+      "step": 200210
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478833762395845e-05,
+      "loss": 0.8364,
+      "step": 200220
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.4787575974530444e-05,
+      "loss": 0.8968,
+      "step": 200230
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478681432510245e-05,
+      "loss": 0.8436,
+      "step": 200240
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478605267567444e-05,
+      "loss": 0.8934,
+      "step": 200250
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478529102624644e-05,
+      "loss": 0.8121,
+      "step": 200260
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.478452937681844e-05,
+      "loss": 0.9402,
+      "step": 200270
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.478376772739044e-05,
+      "loss": 0.938,
+      "step": 200280
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4783006077962436e-05,
+      "loss": 0.8361,
+      "step": 200290
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.478224442853444e-05,
+      "loss": 0.9133,
+      "step": 200300
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.478148277910643e-05,
+      "loss": 0.7986,
+      "step": 200310
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4780721129678435e-05,
+      "loss": 0.8653,
+      "step": 200320
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477995948025043e-05,
+      "loss": 0.8263,
+      "step": 200330
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477919783082243e-05,
+      "loss": 0.9549,
+      "step": 200340
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477843618139443e-05,
+      "loss": 0.8997,
+      "step": 200350
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4777674531966425e-05,
+      "loss": 0.7859,
+      "step": 200360
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477691288253843e-05,
+      "loss": 0.965,
+      "step": 200370
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4776151233110426e-05,
+      "loss": 0.8998,
+      "step": 200380
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4775389583682424e-05,
+      "loss": 0.8138,
+      "step": 200390
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477462793425442e-05,
+      "loss": 0.8401,
+      "step": 200400
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4773866284826426e-05,
+      "loss": 0.8473,
+      "step": 200410
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4773104635398416e-05,
+      "loss": 0.8525,
+      "step": 200420
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477234298597042e-05,
+      "loss": 0.8406,
+      "step": 200430
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477158133654242e-05,
+      "loss": 0.9205,
+      "step": 200440
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.4770819687114416e-05,
+      "loss": 0.8307,
+      "step": 200450
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.477005803768641e-05,
+      "loss": 0.8495,
+      "step": 200460
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.476929638825842e-05,
+      "loss": 0.8069,
+      "step": 200470
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.476853473883041e-05,
+      "loss": 0.9318,
+      "step": 200480
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.476777308940241e-05,
+      "loss": 0.818,
+      "step": 200490
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.476701143997441e-05,
+      "loss": 0.8318,
+      "step": 200500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.476624979054641e-05,
+      "loss": 0.9279,
+      "step": 200510
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4765488141118405e-05,
+      "loss": 0.8114,
+      "step": 200520
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.476472649169041e-05,
+      "loss": 0.8468,
+      "step": 200530
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4763964842262406e-05,
+      "loss": 0.8652,
+      "step": 200540
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4763203192834404e-05,
+      "loss": 0.985,
+      "step": 200550
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.47624415434064e-05,
+      "loss": 0.9415,
+      "step": 200560
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.47616798939784e-05,
+      "loss": 0.855,
+      "step": 200570
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.47609182445504e-05,
+      "loss": 0.852,
+      "step": 200580
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.47601565951224e-05,
+      "loss": 0.8737,
+      "step": 200590
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.47593949456944e-05,
+      "loss": 0.832,
+      "step": 200600
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4758633296266395e-05,
+      "loss": 0.8966,
+      "step": 200610
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.47578716468384e-05,
+      "loss": 0.823,
+      "step": 200620
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.475710999741039e-05,
+      "loss": 0.7928,
+      "step": 200630
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4756348347982394e-05,
+      "loss": 0.8728,
+      "step": 200640
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.475558669855439e-05,
+      "loss": 0.8267,
+      "step": 200650
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.475482504912639e-05,
+      "loss": 0.8681,
+      "step": 200660
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.475406339969839e-05,
+      "loss": 0.8204,
+      "step": 200670
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.475330175027039e-05,
+      "loss": 0.9476,
+      "step": 200680
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.475254010084238e-05,
+      "loss": 0.9029,
+      "step": 200690
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4751778451414386e-05,
+      "loss": 0.8487,
+      "step": 200700
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.4751016801986384e-05,
+      "loss": 0.9955,
+      "step": 200710
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.475025515255838e-05,
+      "loss": 0.7835,
+      "step": 200720
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474949350313038e-05,
+      "loss": 0.9053,
+      "step": 200730
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474873185370238e-05,
+      "loss": 0.8441,
+      "step": 200740
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474797020427438e-05,
+      "loss": 0.7749,
+      "step": 200750
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474720855484638e-05,
+      "loss": 0.919,
+      "step": 200760
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.4746446905418375e-05,
+      "loss": 0.8016,
+      "step": 200770
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474568525599037e-05,
+      "loss": 0.9082,
+      "step": 200780
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474492360656238e-05,
+      "loss": 0.846,
+      "step": 200790
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.4744161957134374e-05,
+      "loss": 0.8538,
+      "step": 200800
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474340030770637e-05,
+      "loss": 0.8584,
+      "step": 200810
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474263865827837e-05,
+      "loss": 0.8061,
+      "step": 200820
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474187700885037e-05,
+      "loss": 0.8874,
+      "step": 200830
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.4741115359422364e-05,
+      "loss": 0.8575,
+      "step": 200840
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.474035370999437e-05,
+      "loss": 0.768,
+      "step": 200850
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.473959206056636e-05,
+      "loss": 0.8659,
+      "step": 200860
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.473883041113836e-05,
+      "loss": 0.9237,
+      "step": 200870
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.473806876171036e-05,
+      "loss": 0.964,
+      "step": 200880
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.473730711228236e-05,
+      "loss": 0.7713,
+      "step": 200890
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.4736545462854356e-05,
+      "loss": 0.9078,
+      "step": 200900
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.473578381342636e-05,
+      "loss": 1.0333,
+      "step": 200910
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.473502216399836e-05,
+      "loss": 0.9612,
+      "step": 200920
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.4734260514570355e-05,
+      "loss": 0.8087,
+      "step": 200930
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.473349886514235e-05,
+      "loss": 0.8176,
+      "step": 200940
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.473273721571435e-05,
+      "loss": 0.8471,
+      "step": 200950
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.4731975566286354e-05,
+      "loss": 0.8012,
+      "step": 200960
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.473121391685835e-05,
+      "loss": 0.8439,
+      "step": 200970
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.473045226743035e-05,
+      "loss": 0.9045,
+      "step": 200980
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.4729690618002346e-05,
+      "loss": 0.8459,
+      "step": 200990
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472892896857435e-05,
+      "loss": 0.8839,
+      "step": 201000
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472816731914634e-05,
+      "loss": 0.8765,
+      "step": 201010
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.4727405669718346e-05,
+      "loss": 0.8849,
+      "step": 201020
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472664402029034e-05,
+      "loss": 0.9204,
+      "step": 201030
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472588237086234e-05,
+      "loss": 0.8233,
+      "step": 201040
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472512072143434e-05,
+      "loss": 0.9267,
+      "step": 201050
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472435907200634e-05,
+      "loss": 0.9098,
+      "step": 201060
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472359742257833e-05,
+      "loss": 0.9256,
+      "step": 201070
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472283577315034e-05,
+      "loss": 0.9823,
+      "step": 201080
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.4722074123722335e-05,
+      "loss": 0.9472,
+      "step": 201090
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472131247429433e-05,
+      "loss": 0.82,
+      "step": 201100
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.472055082486633e-05,
+      "loss": 0.8601,
+      "step": 201110
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.4719789175438334e-05,
+      "loss": 0.8457,
+      "step": 201120
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.471902752601033e-05,
+      "loss": 0.9365,
+      "step": 201130
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.471826587658233e-05,
+      "loss": 0.8382,
+      "step": 201140
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4717504227154326e-05,
+      "loss": 0.9671,
+      "step": 201150
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4716742577726324e-05,
+      "loss": 0.8263,
+      "step": 201160
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.471598092829833e-05,
+      "loss": 0.8778,
+      "step": 201170
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4715219278870325e-05,
+      "loss": 0.901,
+      "step": 201180
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.471445762944232e-05,
+      "loss": 1.0017,
+      "step": 201190
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.471369598001432e-05,
+      "loss": 0.8975,
+      "step": 201200
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4712934330586325e-05,
+      "loss": 0.9229,
+      "step": 201210
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4712172681158315e-05,
+      "loss": 0.8797,
+      "step": 201220
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.471141103173032e-05,
+      "loss": 0.889,
+      "step": 201230
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.471064938230232e-05,
+      "loss": 0.793,
+      "step": 201240
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4709887732874314e-05,
+      "loss": 0.8613,
+      "step": 201250
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.470912608344631e-05,
+      "loss": 0.8927,
+      "step": 201260
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.470836443401831e-05,
+      "loss": 0.8261,
+      "step": 201270
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.470760278459031e-05,
+      "loss": 0.8839,
+      "step": 201280
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.470684113516231e-05,
+      "loss": 0.872,
+      "step": 201290
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.470607948573431e-05,
+      "loss": 0.9252,
+      "step": 201300
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4705317836306306e-05,
+      "loss": 0.9038,
+      "step": 201310
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4704556186878304e-05,
+      "loss": 0.8799,
+      "step": 201320
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.47037945374503e-05,
+      "loss": 0.8065,
+      "step": 201330
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.4703032888022305e-05,
+      "loss": 0.8806,
+      "step": 201340
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.47022712385943e-05,
+      "loss": 0.9049,
+      "step": 201350
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.47015095891663e-05,
+      "loss": 0.8826,
+      "step": 201360
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.47007479397383e-05,
+      "loss": 0.9011,
+      "step": 201370
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.46999862903103e-05,
+      "loss": 0.8639,
+      "step": 201380
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469922464088229e-05,
+      "loss": 0.8331,
+      "step": 201390
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.46984629914543e-05,
+      "loss": 0.8967,
+      "step": 201400
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4697701342026294e-05,
+      "loss": 0.8258,
+      "step": 201410
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469693969259829e-05,
+      "loss": 0.9295,
+      "step": 201420
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469617804317029e-05,
+      "loss": 0.8498,
+      "step": 201430
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4695416393742293e-05,
+      "loss": 0.7737,
+      "step": 201440
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4694654744314284e-05,
+      "loss": 0.876,
+      "step": 201450
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469389309488629e-05,
+      "loss": 0.8919,
+      "step": 201460
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4693131445458286e-05,
+      "loss": 0.9194,
+      "step": 201470
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469236979603028e-05,
+      "loss": 0.8885,
+      "step": 201480
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469160814660228e-05,
+      "loss": 0.8015,
+      "step": 201490
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4690846497174285e-05,
+      "loss": 0.8602,
+      "step": 201500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.469008484774628e-05,
+      "loss": 0.8802,
+      "step": 201510
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.468932319831828e-05,
+      "loss": 0.7963,
+      "step": 201520
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.468856154889028e-05,
+      "loss": 0.8526,
+      "step": 201530
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4687799899462275e-05,
+      "loss": 0.8928,
+      "step": 201540
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.468703825003428e-05,
+      "loss": 0.8305,
+      "step": 201550
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.468627660060628e-05,
+      "loss": 0.8027,
+      "step": 201560
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.4685514951178274e-05,
+      "loss": 0.8246,
+      "step": 201570
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.468475330175027e-05,
+      "loss": 0.875,
+      "step": 201580
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4683991652322276e-05,
+      "loss": 0.8884,
+      "step": 201590
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4683230002894266e-05,
+      "loss": 0.8527,
+      "step": 201600
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.468246835346627e-05,
+      "loss": 0.9058,
+      "step": 201610
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.468170670403827e-05,
+      "loss": 0.8392,
+      "step": 201620
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4680945054610266e-05,
+      "loss": 0.9065,
+      "step": 201630
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.468018340518226e-05,
+      "loss": 0.8814,
+      "step": 201640
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467942175575427e-05,
+      "loss": 0.8482,
+      "step": 201650
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467866010632626e-05,
+      "loss": 0.9081,
+      "step": 201660
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467789845689826e-05,
+      "loss": 0.9859,
+      "step": 201670
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467713680747026e-05,
+      "loss": 0.8682,
+      "step": 201680
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467637515804226e-05,
+      "loss": 0.8511,
+      "step": 201690
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4675613508614255e-05,
+      "loss": 0.919,
+      "step": 201700
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467485185918626e-05,
+      "loss": 0.8512,
+      "step": 201710
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4674090209758256e-05,
+      "loss": 0.8063,
+      "step": 201720
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4673328560330254e-05,
+      "loss": 0.8836,
+      "step": 201730
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467256691090225e-05,
+      "loss": 0.8336,
+      "step": 201740
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467180526147425e-05,
+      "loss": 0.8588,
+      "step": 201750
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.467104361204625e-05,
+      "loss": 0.827,
+      "step": 201760
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4670281962618244e-05,
+      "loss": 0.9507,
+      "step": 201770
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.466952031319025e-05,
+      "loss": 0.8619,
+      "step": 201780
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.4668758663762245e-05,
+      "loss": 0.879,
+      "step": 201790
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.466799701433424e-05,
+      "loss": 0.726,
+      "step": 201800
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.466723536490624e-05,
+      "loss": 0.8869,
+      "step": 201810
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4666473715478245e-05,
+      "loss": 0.9445,
+      "step": 201820
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4665712066050235e-05,
+      "loss": 0.9593,
+      "step": 201830
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.466495041662224e-05,
+      "loss": 0.8833,
+      "step": 201840
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.466418876719424e-05,
+      "loss": 0.8502,
+      "step": 201850
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4663427117766235e-05,
+      "loss": 0.9596,
+      "step": 201860
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.466266546833823e-05,
+      "loss": 0.8549,
+      "step": 201870
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4661903818910236e-05,
+      "loss": 0.8461,
+      "step": 201880
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.466114216948223e-05,
+      "loss": 0.84,
+      "step": 201890
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.466038052005423e-05,
+      "loss": 0.8113,
+      "step": 201900
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465961887062623e-05,
+      "loss": 0.904,
+      "step": 201910
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4658857221198226e-05,
+      "loss": 0.9002,
+      "step": 201920
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465809557177023e-05,
+      "loss": 0.8448,
+      "step": 201930
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465733392234223e-05,
+      "loss": 0.9101,
+      "step": 201940
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4656572272914225e-05,
+      "loss": 0.9073,
+      "step": 201950
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465581062348622e-05,
+      "loss": 0.9394,
+      "step": 201960
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465504897405823e-05,
+      "loss": 0.9409,
+      "step": 201970
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465428732463022e-05,
+      "loss": 0.8535,
+      "step": 201980
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465352567520222e-05,
+      "loss": 0.871,
+      "step": 201990
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465276402577422e-05,
+      "loss": 0.8588,
+      "step": 202000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.465200237634622e-05,
+      "loss": 0.863,
+      "step": 202010
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.4651240726918214e-05,
+      "loss": 0.8419,
+      "step": 202020
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.465047907749022e-05,
+      "loss": 0.86,
+      "step": 202030
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.464971742806221e-05,
+      "loss": 0.8642,
+      "step": 202040
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4648955778634213e-05,
+      "loss": 0.9836,
+      "step": 202050
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.464819412920621e-05,
+      "loss": 0.8478,
+      "step": 202060
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.464743247977821e-05,
+      "loss": 0.8573,
+      "step": 202070
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4646670830350206e-05,
+      "loss": 0.9645,
+      "step": 202080
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.464590918092221e-05,
+      "loss": 0.8852,
+      "step": 202090
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.464514753149421e-05,
+      "loss": 0.8919,
+      "step": 202100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4644385882066205e-05,
+      "loss": 0.8871,
+      "step": 202110
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.46436242326382e-05,
+      "loss": 0.8554,
+      "step": 202120
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.46428625832102e-05,
+      "loss": 0.8711,
+      "step": 202130
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4642100933782204e-05,
+      "loss": 0.8085,
+      "step": 202140
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.46413392843542e-05,
+      "loss": 0.7937,
+      "step": 202150
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.46405776349262e-05,
+      "loss": 0.883,
+      "step": 202160
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.46398159854982e-05,
+      "loss": 0.8857,
+      "step": 202170
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.46390543360702e-05,
+      "loss": 0.7789,
+      "step": 202180
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.463829268664219e-05,
+      "loss": 1.007,
+      "step": 202190
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4637531037214196e-05,
+      "loss": 0.8009,
+      "step": 202200
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4636769387786186e-05,
+      "loss": 0.8561,
+      "step": 202210
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.463600773835819e-05,
+      "loss": 0.7784,
+      "step": 202220
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.463524608893019e-05,
+      "loss": 0.8708,
+      "step": 202230
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.4634484439502186e-05,
+      "loss": 0.8701,
+      "step": 202240
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.463372279007418e-05,
+      "loss": 0.819,
+      "step": 202250
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.463296114064619e-05,
+      "loss": 0.9221,
+      "step": 202260
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.463219949121818e-05,
+      "loss": 0.9094,
+      "step": 202270
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.463143784179018e-05,
+      "loss": 0.944,
+      "step": 202280
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.463067619236218e-05,
+      "loss": 0.8083,
+      "step": 202290
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462991454293418e-05,
+      "loss": 0.8144,
+      "step": 202300
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462915289350618e-05,
+      "loss": 0.9812,
+      "step": 202310
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462839124407818e-05,
+      "loss": 0.9613,
+      "step": 202320
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.4627629594650176e-05,
+      "loss": 0.8429,
+      "step": 202330
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.4626867945222174e-05,
+      "loss": 0.8605,
+      "step": 202340
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462610629579418e-05,
+      "loss": 0.8227,
+      "step": 202350
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462534464636617e-05,
+      "loss": 0.9114,
+      "step": 202360
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462458299693817e-05,
+      "loss": 0.8123,
+      "step": 202370
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462382134751017e-05,
+      "loss": 0.8761,
+      "step": 202380
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462305969808217e-05,
+      "loss": 0.8717,
+      "step": 202390
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.4622298048654165e-05,
+      "loss": 0.8302,
+      "step": 202400
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462153639922617e-05,
+      "loss": 0.8512,
+      "step": 202410
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.462077474979816e-05,
+      "loss": 0.8701,
+      "step": 202420
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.4620013100370165e-05,
+      "loss": 0.7765,
+      "step": 202430
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.461925145094216e-05,
+      "loss": 0.8738,
+      "step": 202440
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.461848980151416e-05,
+      "loss": 0.8596,
+      "step": 202450
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.461772815208616e-05,
+      "loss": 0.863,
+      "step": 202460
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461696650265816e-05,
+      "loss": 0.9175,
+      "step": 202470
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461620485323015e-05,
+      "loss": 0.8783,
+      "step": 202480
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4615443203802156e-05,
+      "loss": 0.8741,
+      "step": 202490
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4614681554374154e-05,
+      "loss": 0.7967,
+      "step": 202500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461391990494615e-05,
+      "loss": 0.8949,
+      "step": 202510
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4613158255518155e-05,
+      "loss": 0.9538,
+      "step": 202520
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461239660609015e-05,
+      "loss": 1.0589,
+      "step": 202530
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461163495666215e-05,
+      "loss": 0.8392,
+      "step": 202540
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461087330723415e-05,
+      "loss": 0.8594,
+      "step": 202550
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.461011165780615e-05,
+      "loss": 0.9711,
+      "step": 202560
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460935000837814e-05,
+      "loss": 0.861,
+      "step": 202570
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460858835895015e-05,
+      "loss": 0.8439,
+      "step": 202580
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4607826709522144e-05,
+      "loss": 0.8604,
+      "step": 202590
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460706506009414e-05,
+      "loss": 0.8586,
+      "step": 202600
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460630341066614e-05,
+      "loss": 0.8599,
+      "step": 202610
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4605541761238144e-05,
+      "loss": 0.8077,
+      "step": 202620
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4604780111810134e-05,
+      "loss": 0.8638,
+      "step": 202630
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460401846238214e-05,
+      "loss": 0.7899,
+      "step": 202640
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4603256812954136e-05,
+      "loss": 0.8981,
+      "step": 202650
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.4602495163526133e-05,
+      "loss": 0.8034,
+      "step": 202660
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460173351409813e-05,
+      "loss": 0.8891,
+      "step": 202670
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.460097186467013e-05,
+      "loss": 0.8398,
+      "step": 202680
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.4600210215242126e-05,
+      "loss": 0.7981,
+      "step": 202690
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459944856581413e-05,
+      "loss": 0.883,
+      "step": 202700
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459868691638613e-05,
+      "loss": 0.8452,
+      "step": 202710
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.4597925266958125e-05,
+      "loss": 0.8811,
+      "step": 202720
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459716361753013e-05,
+      "loss": 0.8222,
+      "step": 202730
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459640196810212e-05,
+      "loss": 0.8523,
+      "step": 202740
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.4595640318674124e-05,
+      "loss": 0.8734,
+      "step": 202750
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459487866924612e-05,
+      "loss": 0.8764,
+      "step": 202760
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459411701981812e-05,
+      "loss": 0.8379,
+      "step": 202770
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459335537039012e-05,
+      "loss": 0.89,
+      "step": 202780
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459259372096212e-05,
+      "loss": 0.8878,
+      "step": 202790
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459183207153411e-05,
+      "loss": 0.7851,
+      "step": 202800
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.4591070422106116e-05,
+      "loss": 0.8976,
+      "step": 202810
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.459030877267811e-05,
+      "loss": 0.9046,
+      "step": 202820
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.458954712325011e-05,
+      "loss": 0.8201,
+      "step": 202830
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.458878547382211e-05,
+      "loss": 0.8702,
+      "step": 202840
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.458802382439411e-05,
+      "loss": 0.9096,
+      "step": 202850
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.45872621749661e-05,
+      "loss": 0.7896,
+      "step": 202860
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.458650052553811e-05,
+      "loss": 0.8633,
+      "step": 202870
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.4585738876110105e-05,
+      "loss": 0.8342,
+      "step": 202880
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.45849772266821e-05,
+      "loss": 0.7748,
+      "step": 202890
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.4584215577254107e-05,
+      "loss": 0.9117,
+      "step": 202900
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4583453927826104e-05,
+      "loss": 0.8683,
+      "step": 202910
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.45826922783981e-05,
+      "loss": 0.938,
+      "step": 202920
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.45819306289701e-05,
+      "loss": 0.8981,
+      "step": 202930
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.45811689795421e-05,
+      "loss": 0.9321,
+      "step": 202940
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4580407330114094e-05,
+      "loss": 0.8462,
+      "step": 202950
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.45796456806861e-05,
+      "loss": 0.8438,
+      "step": 202960
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4578884031258096e-05,
+      "loss": 0.8741,
+      "step": 202970
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457812238183009e-05,
+      "loss": 0.7923,
+      "step": 202980
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457736073240209e-05,
+      "loss": 0.8075,
+      "step": 202990
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4576599082974095e-05,
+      "loss": 0.8889,
+      "step": 203000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4575837433546085e-05,
+      "loss": 0.7797,
+      "step": 203010
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457507578411809e-05,
+      "loss": 0.915,
+      "step": 203020
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457431413469009e-05,
+      "loss": 0.8202,
+      "step": 203030
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4573552485262085e-05,
+      "loss": 0.7682,
+      "step": 203040
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457279083583408e-05,
+      "loss": 0.8941,
+      "step": 203050
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4572029186406086e-05,
+      "loss": 0.7973,
+      "step": 203060
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457126753697808e-05,
+      "loss": 0.9843,
+      "step": 203070
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.457050588755008e-05,
+      "loss": 0.8566,
+      "step": 203080
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.456974423812208e-05,
+      "loss": 0.8567,
+      "step": 203090
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.4568982588694076e-05,
+      "loss": 0.7898,
+      "step": 203100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.456822093926608e-05,
+      "loss": 0.8288,
+      "step": 203110
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.456745928983807e-05,
+      "loss": 0.9171,
+      "step": 203120
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4566697640410075e-05,
+      "loss": 0.8557,
+      "step": 203130
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456593599098207e-05,
+      "loss": 0.8845,
+      "step": 203140
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456517434155407e-05,
+      "loss": 0.9312,
+      "step": 203150
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456441269212607e-05,
+      "loss": 0.8463,
+      "step": 203160
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456365104269807e-05,
+      "loss": 0.8454,
+      "step": 203170
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456288939327006e-05,
+      "loss": 0.8505,
+      "step": 203180
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456212774384207e-05,
+      "loss": 0.8008,
+      "step": 203190
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4561366094414064e-05,
+      "loss": 0.8612,
+      "step": 203200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.456060444498606e-05,
+      "loss": 0.9228,
+      "step": 203210
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.455984279555806e-05,
+      "loss": 0.8336,
+      "step": 203220
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4559081146130064e-05,
+      "loss": 0.8414,
+      "step": 203230
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4558319496702054e-05,
+      "loss": 0.7894,
+      "step": 203240
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.455755784727406e-05,
+      "loss": 0.8123,
+      "step": 203250
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4556796197846056e-05,
+      "loss": 0.7962,
+      "step": 203260
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4556034548418053e-05,
+      "loss": 0.9439,
+      "step": 203270
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.455527289899005e-05,
+      "loss": 0.9094,
+      "step": 203280
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4554511249562055e-05,
+      "loss": 0.8783,
+      "step": 203290
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.455374960013405e-05,
+      "loss": 0.9235,
+      "step": 203300
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.455298795070605e-05,
+      "loss": 0.8417,
+      "step": 203310
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4552226301278054e-05,
+      "loss": 0.9236,
+      "step": 203320
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.4551464651850045e-05,
+      "loss": 0.8723,
+      "step": 203330
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.455070300242205e-05,
+      "loss": 0.7764,
+      "step": 203340
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454994135299405e-05,
+      "loss": 0.97,
+      "step": 203350
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.4549179703566044e-05,
+      "loss": 0.9157,
+      "step": 203360
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454841805413804e-05,
+      "loss": 0.8072,
+      "step": 203370
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.4547656404710046e-05,
+      "loss": 0.9183,
+      "step": 203380
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454689475528204e-05,
+      "loss": 0.8669,
+      "step": 203390
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454613310585404e-05,
+      "loss": 0.7924,
+      "step": 203400
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454537145642604e-05,
+      "loss": 0.9347,
+      "step": 203410
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.4544609806998036e-05,
+      "loss": 0.9205,
+      "step": 203420
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454384815757003e-05,
+      "loss": 0.9065,
+      "step": 203430
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454308650814204e-05,
+      "loss": 0.9244,
+      "step": 203440
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454232485871403e-05,
+      "loss": 0.9225,
+      "step": 203450
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454156320928603e-05,
+      "loss": 0.8034,
+      "step": 203460
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454080155985803e-05,
+      "loss": 0.9556,
+      "step": 203470
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.454003991043003e-05,
+      "loss": 0.8616,
+      "step": 203480
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.4539278261002025e-05,
+      "loss": 0.9321,
+      "step": 203490
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.453851661157403e-05,
+      "loss": 0.8019,
+      "step": 203500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.4537754962146027e-05,
+      "loss": 0.8693,
+      "step": 203510
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.4536993312718024e-05,
+      "loss": 0.8346,
+      "step": 203520
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.453623166329003e-05,
+      "loss": 0.8457,
+      "step": 203530
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.453547001386202e-05,
+      "loss": 0.8629,
+      "step": 203540
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.453470836443402e-05,
+      "loss": 0.8447,
+      "step": 203550
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.453394671500602e-05,
+      "loss": 0.7953,
+      "step": 203560
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.453318506557802e-05,
+      "loss": 0.9336,
+      "step": 203570
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4532423416150016e-05,
+      "loss": 0.8517,
+      "step": 203580
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.453166176672201e-05,
+      "loss": 0.8026,
+      "step": 203590
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.453090011729401e-05,
+      "loss": 0.886,
+      "step": 203600
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4530138467866015e-05,
+      "loss": 0.8936,
+      "step": 203610
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4529376818438005e-05,
+      "loss": 0.8547,
+      "step": 203620
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.452861516901001e-05,
+      "loss": 0.846,
+      "step": 203630
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.452785351958201e-05,
+      "loss": 0.886,
+      "step": 203640
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4527091870154005e-05,
+      "loss": 0.757,
+      "step": 203650
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4526330220726e-05,
+      "loss": 0.8497,
+      "step": 203660
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4525568571298006e-05,
+      "loss": 0.8002,
+      "step": 203670
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4524806921870004e-05,
+      "loss": 0.8144,
+      "step": 203680
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4524045272442e-05,
+      "loss": 1.0171,
+      "step": 203690
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4523283623014006e-05,
+      "loss": 0.8639,
+      "step": 203700
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4522521973585996e-05,
+      "loss": 0.9002,
+      "step": 203710
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4521760324158e-05,
+      "loss": 0.897,
+      "step": 203720
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.452099867473e-05,
+      "loss": 0.7711,
+      "step": 203730
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4520237025301995e-05,
+      "loss": 0.933,
+      "step": 203740
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.451947537587399e-05,
+      "loss": 0.8619,
+      "step": 203750
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.4518713726446e-05,
+      "loss": 0.7914,
+      "step": 203760
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.451795207701799e-05,
+      "loss": 0.9393,
+      "step": 203770
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451719042758999e-05,
+      "loss": 0.9676,
+      "step": 203780
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451642877816199e-05,
+      "loss": 0.7853,
+      "step": 203790
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451566712873399e-05,
+      "loss": 0.9224,
+      "step": 203800
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4514905479305984e-05,
+      "loss": 0.7954,
+      "step": 203810
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451414382987799e-05,
+      "loss": 0.763,
+      "step": 203820
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451338218044998e-05,
+      "loss": 0.9677,
+      "step": 203830
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4512620531021984e-05,
+      "loss": 0.8916,
+      "step": 203840
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451185888159398e-05,
+      "loss": 0.8093,
+      "step": 203850
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.451109723216598e-05,
+      "loss": 0.898,
+      "step": 203860
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4510335582737976e-05,
+      "loss": 0.7774,
+      "step": 203870
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450957393330998e-05,
+      "loss": 0.9995,
+      "step": 203880
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450881228388198e-05,
+      "loss": 0.8432,
+      "step": 203890
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4508050634453975e-05,
+      "loss": 0.8548,
+      "step": 203900
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450728898502598e-05,
+      "loss": 0.8324,
+      "step": 203910
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450652733559797e-05,
+      "loss": 0.8576,
+      "step": 203920
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4505765686169974e-05,
+      "loss": 0.8525,
+      "step": 203930
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450500403674197e-05,
+      "loss": 0.8305,
+      "step": 203940
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450424238731397e-05,
+      "loss": 0.8243,
+      "step": 203950
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450348073788597e-05,
+      "loss": 0.8419,
+      "step": 203960
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450271908845797e-05,
+      "loss": 0.8108,
+      "step": 203970
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.450195743902996e-05,
+      "loss": 0.9438,
+      "step": 203980
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4501195789601966e-05,
+      "loss": 0.8348,
+      "step": 203990
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4500434140173963e-05,
+      "loss": 0.869,
+      "step": 204000
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449967249074596e-05,
+      "loss": 0.9382,
+      "step": 204010
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449891084131796e-05,
+      "loss": 0.8993,
+      "step": 204020
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449814919188996e-05,
+      "loss": 0.8071,
+      "step": 204030
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449738754246195e-05,
+      "loss": 0.8336,
+      "step": 204040
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449662589303396e-05,
+      "loss": 0.8776,
+      "step": 204050
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4495864243605955e-05,
+      "loss": 0.915,
+      "step": 204060
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449510259417795e-05,
+      "loss": 0.9058,
+      "step": 204070
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449434094474995e-05,
+      "loss": 0.853,
+      "step": 204080
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449357929532195e-05,
+      "loss": 0.8109,
+      "step": 204090
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449281764589395e-05,
+      "loss": 0.7953,
+      "step": 204100
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.449205599646595e-05,
+      "loss": 0.8612,
+      "step": 204110
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4491294347037947e-05,
+      "loss": 0.8667,
+      "step": 204120
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4490532697609944e-05,
+      "loss": 0.7956,
+      "step": 204130
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448977104818195e-05,
+      "loss": 0.7231,
+      "step": 204140
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448900939875394e-05,
+      "loss": 0.8818,
+      "step": 204150
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448824774932594e-05,
+      "loss": 0.8577,
+      "step": 204160
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448748609989794e-05,
+      "loss": 0.965,
+      "step": 204170
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448672445046994e-05,
+      "loss": 0.9214,
+      "step": 204180
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4485962801041936e-05,
+      "loss": 0.9092,
+      "step": 204190
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448520115161394e-05,
+      "loss": 0.7637,
+      "step": 204200
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.448443950218593e-05,
+      "loss": 0.8647,
+      "step": 204210
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4483677852757935e-05,
+      "loss": 0.8343,
+      "step": 204220
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.448291620332993e-05,
+      "loss": 0.8186,
+      "step": 204230
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.448215455390193e-05,
+      "loss": 0.8683,
+      "step": 204240
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.448139290447393e-05,
+      "loss": 0.7494,
+      "step": 204250
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.448063125504593e-05,
+      "loss": 0.8106,
+      "step": 204260
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447986960561793e-05,
+      "loss": 0.8833,
+      "step": 204270
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4479107956189926e-05,
+      "loss": 0.8884,
+      "step": 204280
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4478346306761924e-05,
+      "loss": 0.8985,
+      "step": 204290
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447758465733392e-05,
+      "loss": 0.8525,
+      "step": 204300
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4476823007905926e-05,
+      "loss": 0.8221,
+      "step": 204310
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447606135847792e-05,
+      "loss": 0.882,
+      "step": 204320
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447529970904992e-05,
+      "loss": 0.8845,
+      "step": 204330
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447453805962192e-05,
+      "loss": 0.8291,
+      "step": 204340
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447377641019392e-05,
+      "loss": 0.8036,
+      "step": 204350
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447301476076591e-05,
+      "loss": 0.9268,
+      "step": 204360
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447225311133792e-05,
+      "loss": 0.8017,
+      "step": 204370
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4471491461909915e-05,
+      "loss": 0.8304,
+      "step": 204380
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.447072981248191e-05,
+      "loss": 0.8584,
+      "step": 204390
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.446996816305391e-05,
+      "loss": 0.7904,
+      "step": 204400
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4469206513625914e-05,
+      "loss": 0.7977,
+      "step": 204410
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.4468444864197904e-05,
+      "loss": 0.9039,
+      "step": 204420
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.446768321476991e-05,
+      "loss": 0.8423,
+      "step": 204430
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4466921565341906e-05,
+      "loss": 0.8126,
+      "step": 204440
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4466159915913904e-05,
+      "loss": 0.793,
+      "step": 204450
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.44653982664859e-05,
+      "loss": 0.9211,
+      "step": 204460
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4464636617057905e-05,
+      "loss": 1.0471,
+      "step": 204470
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.44638749676299e-05,
+      "loss": 0.9072,
+      "step": 204480
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.44631133182019e-05,
+      "loss": 0.9169,
+      "step": 204490
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.44623516687739e-05,
+      "loss": 0.968,
+      "step": 204500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4461590019345895e-05,
+      "loss": 0.8236,
+      "step": 204510
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.44608283699179e-05,
+      "loss": 0.888,
+      "step": 204520
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.446006672048989e-05,
+      "loss": 0.8118,
+      "step": 204530
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4459305071061894e-05,
+      "loss": 0.8579,
+      "step": 204540
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445854342163389e-05,
+      "loss": 0.822,
+      "step": 204550
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445778177220589e-05,
+      "loss": 0.8851,
+      "step": 204560
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445702012277789e-05,
+      "loss": 0.8623,
+      "step": 204570
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445625847334989e-05,
+      "loss": 0.7322,
+      "step": 204580
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445549682392188e-05,
+      "loss": 0.8058,
+      "step": 204590
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4454735174493886e-05,
+      "loss": 0.9124,
+      "step": 204600
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.4453973525065883e-05,
+      "loss": 0.9188,
+      "step": 204610
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445321187563788e-05,
+      "loss": 0.8496,
+      "step": 204620
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445245022620988e-05,
+      "loss": 0.8447,
+      "step": 204630
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445168857678188e-05,
+      "loss": 0.8822,
+      "step": 204640
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.445092692735388e-05,
+      "loss": 0.9229,
+      "step": 204650
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.445016527792588e-05,
+      "loss": 0.8886,
+      "step": 204660
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4449403628497875e-05,
+      "loss": 0.885,
+      "step": 204670
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444864197906987e-05,
+      "loss": 0.8667,
+      "step": 204680
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444788032964188e-05,
+      "loss": 0.9407,
+      "step": 204690
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4447118680213874e-05,
+      "loss": 0.8099,
+      "step": 204700
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444635703078587e-05,
+      "loss": 0.8282,
+      "step": 204710
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444559538135787e-05,
+      "loss": 0.8385,
+      "step": 204720
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444483373192987e-05,
+      "loss": 0.9053,
+      "step": 204730
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4444072082501864e-05,
+      "loss": 0.7725,
+      "step": 204740
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444331043307387e-05,
+      "loss": 0.8279,
+      "step": 204750
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4442548783645866e-05,
+      "loss": 0.8967,
+      "step": 204760
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444178713421786e-05,
+      "loss": 0.8141,
+      "step": 204770
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.444102548478986e-05,
+      "loss": 0.8992,
+      "step": 204780
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4440263835361865e-05,
+      "loss": 0.9038,
+      "step": 204790
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4439502185933856e-05,
+      "loss": 0.8586,
+      "step": 204800
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.443874053650586e-05,
+      "loss": 0.9467,
+      "step": 204810
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.443797888707786e-05,
+      "loss": 0.9362,
+      "step": 204820
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4437217237649855e-05,
+      "loss": 0.8907,
+      "step": 204830
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.443645558822185e-05,
+      "loss": 0.8535,
+      "step": 204840
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4435693938793857e-05,
+      "loss": 0.8229,
+      "step": 204850
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.4434932289365854e-05,
+      "loss": 0.9175,
+      "step": 204860
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.443417063993785e-05,
+      "loss": 0.8856,
+      "step": 204870
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.443340899050985e-05,
+      "loss": 0.8784,
+      "step": 204880
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4432647341081846e-05,
+      "loss": 0.8255,
+      "step": 204890
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.443188569165385e-05,
+      "loss": 0.7902,
+      "step": 204900
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.443112404222585e-05,
+      "loss": 0.7834,
+      "step": 204910
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4430362392797846e-05,
+      "loss": 0.7882,
+      "step": 204920
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442960074336984e-05,
+      "loss": 0.8817,
+      "step": 204930
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442883909394185e-05,
+      "loss": 0.9359,
+      "step": 204940
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442807744451384e-05,
+      "loss": 0.8501,
+      "step": 204950
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442731579508584e-05,
+      "loss": 0.9703,
+      "step": 204960
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442655414565783e-05,
+      "loss": 0.905,
+      "step": 204970
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442579249622984e-05,
+      "loss": 0.9578,
+      "step": 204980
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4425030846801835e-05,
+      "loss": 0.8952,
+      "step": 204990
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442426919737383e-05,
+      "loss": 0.9203,
+      "step": 205000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442350754794583e-05,
+      "loss": 0.9437,
+      "step": 205010
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4422745898517834e-05,
+      "loss": 0.7561,
+      "step": 205020
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442198424908983e-05,
+      "loss": 0.921,
+      "step": 205030
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.442122259966183e-05,
+      "loss": 0.8285,
+      "step": 205040
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4420460950233826e-05,
+      "loss": 0.8599,
+      "step": 205050
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4419699300805824e-05,
+      "loss": 0.7818,
+      "step": 205060
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.441893765137783e-05,
+      "loss": 0.876,
+      "step": 205070
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.4418176001949825e-05,
+      "loss": 0.7319,
+      "step": 205080
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.441741435252182e-05,
+      "loss": 0.8225,
+      "step": 205090
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.441665270309382e-05,
+      "loss": 0.8841,
+      "step": 205100
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4415891053665825e-05,
+      "loss": 0.9384,
+      "step": 205110
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4415129404237815e-05,
+      "loss": 0.9655,
+      "step": 205120
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.441436775480982e-05,
+      "loss": 0.8558,
+      "step": 205130
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.441360610538182e-05,
+      "loss": 0.8058,
+      "step": 205140
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4412844455953814e-05,
+      "loss": 0.8857,
+      "step": 205150
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.441208280652581e-05,
+      "loss": 0.9282,
+      "step": 205160
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4411321157097816e-05,
+      "loss": 0.9793,
+      "step": 205170
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.441055950766981e-05,
+      "loss": 1.0046,
+      "step": 205180
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.440979785824181e-05,
+      "loss": 0.8237,
+      "step": 205190
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.440903620881381e-05,
+      "loss": 0.8606,
+      "step": 205200
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4408274559385806e-05,
+      "loss": 0.8657,
+      "step": 205210
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4407512909957803e-05,
+      "loss": 0.8944,
+      "step": 205220
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.440675126052981e-05,
+      "loss": 0.9638,
+      "step": 205230
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4405989611101805e-05,
+      "loss": 0.9273,
+      "step": 205240
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.44052279616738e-05,
+      "loss": 0.8562,
+      "step": 205250
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.44044663122458e-05,
+      "loss": 0.7885,
+      "step": 205260
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.44037046628178e-05,
+      "loss": 0.8399,
+      "step": 205270
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.44029430133898e-05,
+      "loss": 0.8723,
+      "step": 205280
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.44021813639618e-05,
+      "loss": 0.8489,
+      "step": 205290
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.44014197145338e-05,
+      "loss": 0.8127,
+      "step": 205300
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.4400658065105794e-05,
+      "loss": 0.8432,
+      "step": 205310
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.43998964156778e-05,
+      "loss": 0.8195,
+      "step": 205320
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439913476624979e-05,
+      "loss": 0.8553,
+      "step": 205330
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439837311682179e-05,
+      "loss": 0.8586,
+      "step": 205340
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439761146739379e-05,
+      "loss": 0.7427,
+      "step": 205350
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439684981796579e-05,
+      "loss": 0.8198,
+      "step": 205360
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4396088168537786e-05,
+      "loss": 0.9395,
+      "step": 205370
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439532651910979e-05,
+      "loss": 0.9698,
+      "step": 205380
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439456486968178e-05,
+      "loss": 0.843,
+      "step": 205390
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4393803220253785e-05,
+      "loss": 0.816,
+      "step": 205400
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439304157082578e-05,
+      "loss": 0.8424,
+      "step": 205410
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439227992139778e-05,
+      "loss": 0.9451,
+      "step": 205420
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.439151827196978e-05,
+      "loss": 0.9228,
+      "step": 205430
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4390756622541775e-05,
+      "loss": 0.9122,
+      "step": 205440
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.438999497311378e-05,
+      "loss": 0.7943,
+      "step": 205450
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4389233323685777e-05,
+      "loss": 0.9196,
+      "step": 205460
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4388471674257774e-05,
+      "loss": 0.8764,
+      "step": 205470
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.438771002482977e-05,
+      "loss": 0.8317,
+      "step": 205480
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4386948375401776e-05,
+      "loss": 0.9321,
+      "step": 205490
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.4386186725973766e-05,
+      "loss": 0.894,
+      "step": 205500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.438542507654577e-05,
+      "loss": 0.8188,
+      "step": 205510
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.438466342711777e-05,
+      "loss": 0.9128,
+      "step": 205520
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.4383901777689766e-05,
+      "loss": 0.8652,
+      "step": 205530
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.438314012826176e-05,
+      "loss": 0.8416,
+      "step": 205540
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.438237847883377e-05,
+      "loss": 0.802,
+      "step": 205550
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.438161682940576e-05,
+      "loss": 0.8465,
+      "step": 205560
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.438085517997776e-05,
+      "loss": 0.9186,
+      "step": 205570
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.438009353054976e-05,
+      "loss": 0.8568,
+      "step": 205580
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437933188112176e-05,
+      "loss": 0.9248,
+      "step": 205590
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.4378570231693755e-05,
+      "loss": 0.9697,
+      "step": 205600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437780858226576e-05,
+      "loss": 0.8178,
+      "step": 205610
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.4377046932837756e-05,
+      "loss": 0.7503,
+      "step": 205620
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.4376285283409754e-05,
+      "loss": 0.8471,
+      "step": 205630
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437552363398175e-05,
+      "loss": 0.8172,
+      "step": 205640
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437476198455375e-05,
+      "loss": 1.0345,
+      "step": 205650
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437400033512575e-05,
+      "loss": 0.9083,
+      "step": 205660
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437323868569775e-05,
+      "loss": 0.882,
+      "step": 205670
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437247703626975e-05,
+      "loss": 0.8833,
+      "step": 205680
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.4371715386841745e-05,
+      "loss": 0.7691,
+      "step": 205690
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437095373741375e-05,
+      "loss": 0.8822,
+      "step": 205700
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.437019208798574e-05,
+      "loss": 0.8708,
+      "step": 205710
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.4369430438557745e-05,
+      "loss": 0.7822,
+      "step": 205720
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.436866878912974e-05,
+      "loss": 0.7831,
+      "step": 205730
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.436790713970174e-05,
+      "loss": 0.8177,
+      "step": 205740
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436714549027374e-05,
+      "loss": 0.9408,
+      "step": 205750
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436638384084574e-05,
+      "loss": 0.9075,
+      "step": 205760
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436562219141773e-05,
+      "loss": 0.9315,
+      "step": 205770
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.4364860541989736e-05,
+      "loss": 0.8324,
+      "step": 205780
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.4364098892561734e-05,
+      "loss": 0.847,
+      "step": 205790
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436333724313373e-05,
+      "loss": 0.7956,
+      "step": 205800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436257559370573e-05,
+      "loss": 0.8442,
+      "step": 205810
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436181394427773e-05,
+      "loss": 0.7904,
+      "step": 205820
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436105229484973e-05,
+      "loss": 0.816,
+      "step": 205830
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.436029064542173e-05,
+      "loss": 0.8903,
+      "step": 205840
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.4359528995993725e-05,
+      "loss": 0.9178,
+      "step": 205850
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435876734656572e-05,
+      "loss": 0.816,
+      "step": 205860
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435800569713773e-05,
+      "loss": 0.8681,
+      "step": 205870
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435724404770972e-05,
+      "loss": 0.8881,
+      "step": 205880
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435648239828172e-05,
+      "loss": 0.8574,
+      "step": 205890
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435572074885372e-05,
+      "loss": 0.799,
+      "step": 205900
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435495909942572e-05,
+      "loss": 0.9273,
+      "step": 205910
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.4354197449997714e-05,
+      "loss": 0.8212,
+      "step": 205920
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435343580056972e-05,
+      "loss": 0.8127,
+      "step": 205930
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435267415114171e-05,
+      "loss": 0.865,
+      "step": 205940
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.4351912501713713e-05,
+      "loss": 0.9241,
+      "step": 205950
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.435115085228571e-05,
+      "loss": 0.9375,
+      "step": 205960
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.435038920285771e-05,
+      "loss": 0.8609,
+      "step": 205970
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.4349627553429706e-05,
+      "loss": 0.8845,
+      "step": 205980
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.434886590400171e-05,
+      "loss": 0.8477,
+      "step": 205990
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.434810425457371e-05,
+      "loss": 0.9128,
+      "step": 206000
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.4347342605145705e-05,
+      "loss": 0.848,
+      "step": 206010
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.43465809557177e-05,
+      "loss": 0.8836,
+      "step": 206020
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.43458193062897e-05,
+      "loss": 0.9721,
+      "step": 206030
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.4345057656861704e-05,
+      "loss": 0.9596,
+      "step": 206040
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.43442960074337e-05,
+      "loss": 0.8594,
+      "step": 206050
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.43435343580057e-05,
+      "loss": 0.8436,
+      "step": 206060
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.4342772708577697e-05,
+      "loss": 0.9249,
+      "step": 206070
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.43420110591497e-05,
+      "loss": 0.7962,
+      "step": 206080
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.434124940972169e-05,
+      "loss": 0.776,
+      "step": 206090
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.4340487760293696e-05,
+      "loss": 0.8293,
+      "step": 206100
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433972611086569e-05,
+      "loss": 0.7956,
+      "step": 206110
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433896446143769e-05,
+      "loss": 0.8615,
+      "step": 206120
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433820281200969e-05,
+      "loss": 1.0254,
+      "step": 206130
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433744116258169e-05,
+      "loss": 0.8079,
+      "step": 206140
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433667951315368e-05,
+      "loss": 0.8602,
+      "step": 206150
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433591786372569e-05,
+      "loss": 0.8539,
+      "step": 206160
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.4335156214297685e-05,
+      "loss": 0.8174,
+      "step": 206170
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.433439456486968e-05,
+      "loss": 0.8421,
+      "step": 206180
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.433363291544168e-05,
+      "loss": 0.7804,
+      "step": 206190
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4332871266013684e-05,
+      "loss": 0.921,
+      "step": 206200
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.433210961658568e-05,
+      "loss": 0.8435,
+      "step": 206210
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.433134796715768e-05,
+      "loss": 0.9111,
+      "step": 206220
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4330586317729676e-05,
+      "loss": 0.8386,
+      "step": 206230
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4329824668301674e-05,
+      "loss": 0.9259,
+      "step": 206240
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432906301887368e-05,
+      "loss": 0.8498,
+      "step": 206250
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4328301369445676e-05,
+      "loss": 0.8144,
+      "step": 206260
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432753972001767e-05,
+      "loss": 0.7662,
+      "step": 206270
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432677807058967e-05,
+      "loss": 0.995,
+      "step": 206280
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4326016421161675e-05,
+      "loss": 0.8696,
+      "step": 206290
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4325254771733665e-05,
+      "loss": 0.8803,
+      "step": 206300
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432449312230567e-05,
+      "loss": 0.7295,
+      "step": 206310
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432373147287767e-05,
+      "loss": 0.8841,
+      "step": 206320
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4322969823449665e-05,
+      "loss": 0.7738,
+      "step": 206330
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432220817402166e-05,
+      "loss": 0.8501,
+      "step": 206340
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432144652459366e-05,
+      "loss": 0.9398,
+      "step": 206350
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.432068487516566e-05,
+      "loss": 0.8249,
+      "step": 206360
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.431992322573766e-05,
+      "loss": 0.9078,
+      "step": 206370
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.431916157630965e-05,
+      "loss": 0.8266,
+      "step": 206380
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4318399926881656e-05,
+      "loss": 0.8257,
+      "step": 206390
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.4317638277453654e-05,
+      "loss": 0.9205,
+      "step": 206400
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431687662802565e-05,
+      "loss": 0.8447,
+      "step": 206410
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4316114978597655e-05,
+      "loss": 0.86,
+      "step": 206420
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431535332916965e-05,
+      "loss": 0.8839,
+      "step": 206430
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431459167974165e-05,
+      "loss": 0.9424,
+      "step": 206440
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431383003031365e-05,
+      "loss": 1.034,
+      "step": 206450
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431306838088565e-05,
+      "loss": 0.844,
+      "step": 206460
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431230673145764e-05,
+      "loss": 0.7589,
+      "step": 206470
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431154508202965e-05,
+      "loss": 0.9089,
+      "step": 206480
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4310783432601644e-05,
+      "loss": 0.8972,
+      "step": 206490
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.431002178317364e-05,
+      "loss": 0.868,
+      "step": 206500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.430926013374564e-05,
+      "loss": 0.8505,
+      "step": 206510
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4308498484317644e-05,
+      "loss": 0.7948,
+      "step": 206520
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4307736834889634e-05,
+      "loss": 0.8624,
+      "step": 206530
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.430697518546164e-05,
+      "loss": 0.8326,
+      "step": 206540
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4306213536033636e-05,
+      "loss": 0.8731,
+      "step": 206550
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4305451886605633e-05,
+      "loss": 0.875,
+      "step": 206560
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.430469023717763e-05,
+      "loss": 0.9216,
+      "step": 206570
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4303928587749635e-05,
+      "loss": 0.9141,
+      "step": 206580
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.430316693832163e-05,
+      "loss": 0.7581,
+      "step": 206590
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.430240528889363e-05,
+      "loss": 0.853,
+      "step": 206600
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.430164363946563e-05,
+      "loss": 0.7853,
+      "step": 206610
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.4300881990037625e-05,
+      "loss": 0.8443,
+      "step": 206620
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.430012034060963e-05,
+      "loss": 0.7476,
+      "step": 206630
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429935869118163e-05,
+      "loss": 0.9267,
+      "step": 206640
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4298597041753624e-05,
+      "loss": 0.9568,
+      "step": 206650
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429783539232562e-05,
+      "loss": 0.85,
+      "step": 206660
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4297073742897626e-05,
+      "loss": 0.8863,
+      "step": 206670
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4296312093469617e-05,
+      "loss": 0.9004,
+      "step": 206680
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429555044404162e-05,
+      "loss": 0.7894,
+      "step": 206690
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429478879461362e-05,
+      "loss": 0.837,
+      "step": 206700
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4294027145185616e-05,
+      "loss": 0.8196,
+      "step": 206710
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429326549575761e-05,
+      "loss": 0.8399,
+      "step": 206720
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429250384632962e-05,
+      "loss": 0.8849,
+      "step": 206730
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429174219690161e-05,
+      "loss": 0.9135,
+      "step": 206740
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429098054747361e-05,
+      "loss": 0.9253,
+      "step": 206750
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.429021889804561e-05,
+      "loss": 0.9112,
+      "step": 206760
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.428945724861761e-05,
+      "loss": 0.8958,
+      "step": 206770
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4288695599189605e-05,
+      "loss": 0.9763,
+      "step": 206780
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.428793394976161e-05,
+      "loss": 0.8439,
+      "step": 206790
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4287172300333606e-05,
+      "loss": 0.9102,
+      "step": 206800
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.4286410650905604e-05,
+      "loss": 0.8965,
+      "step": 206810
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.42856490014776e-05,
+      "loss": 0.8743,
+      "step": 206820
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.42848873520496e-05,
+      "loss": 0.8663,
+      "step": 206830
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.42841257026216e-05,
+      "loss": 0.8912,
+      "step": 206840
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4283364053193594e-05,
+      "loss": 0.9034,
+      "step": 206850
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.42826024037656e-05,
+      "loss": 0.8323,
+      "step": 206860
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4281840754337596e-05,
+      "loss": 0.8229,
+      "step": 206870
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.428107910490959e-05,
+      "loss": 0.8616,
+      "step": 206880
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.428031745548159e-05,
+      "loss": 0.8871,
+      "step": 206890
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4279555806053595e-05,
+      "loss": 0.7842,
+      "step": 206900
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4278794156625585e-05,
+      "loss": 0.8079,
+      "step": 206910
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427803250719759e-05,
+      "loss": 0.9284,
+      "step": 206920
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427727085776959e-05,
+      "loss": 0.8941,
+      "step": 206930
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4276509208341585e-05,
+      "loss": 0.8525,
+      "step": 206940
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427574755891358e-05,
+      "loss": 0.8565,
+      "step": 206950
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4274985909485586e-05,
+      "loss": 0.9312,
+      "step": 206960
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427422426005758e-05,
+      "loss": 0.8358,
+      "step": 206970
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427346261062958e-05,
+      "loss": 0.9082,
+      "step": 206980
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427270096120158e-05,
+      "loss": 0.9484,
+      "step": 206990
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4271939311773576e-05,
+      "loss": 0.9017,
+      "step": 207000
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427117766234558e-05,
+      "loss": 0.8116,
+      "step": 207010
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.427041601291758e-05,
+      "loss": 0.8017,
+      "step": 207020
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.4269654363489575e-05,
+      "loss": 0.939,
+      "step": 207030
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.426889271406157e-05,
+      "loss": 1.064,
+      "step": 207040
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.426813106463358e-05,
+      "loss": 0.8396,
+      "step": 207050
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.426736941520557e-05,
+      "loss": 0.8176,
+      "step": 207060
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426660776577757e-05,
+      "loss": 0.8864,
+      "step": 207070
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426584611634957e-05,
+      "loss": 0.8851,
+      "step": 207080
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426508446692157e-05,
+      "loss": 0.8322,
+      "step": 207090
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.4264322817493564e-05,
+      "loss": 0.8984,
+      "step": 207100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426356116806557e-05,
+      "loss": 0.7929,
+      "step": 207110
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426279951863756e-05,
+      "loss": 0.7828,
+      "step": 207120
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.4262037869209564e-05,
+      "loss": 0.8481,
+      "step": 207130
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426127621978156e-05,
+      "loss": 0.9301,
+      "step": 207140
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.426051457035356e-05,
+      "loss": 0.9861,
+      "step": 207150
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.4259752920925556e-05,
+      "loss": 0.8036,
+      "step": 207160
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425899127149756e-05,
+      "loss": 0.9043,
+      "step": 207170
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425822962206955e-05,
+      "loss": 0.7978,
+      "step": 207180
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.4257467972641555e-05,
+      "loss": 0.8028,
+      "step": 207190
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425670632321355e-05,
+      "loss": 0.8976,
+      "step": 207200
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425594467378555e-05,
+      "loss": 0.7865,
+      "step": 207210
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.4255183024357554e-05,
+      "loss": 0.8582,
+      "step": 207220
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425442137492955e-05,
+      "loss": 0.84,
+      "step": 207230
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425365972550155e-05,
+      "loss": 0.8229,
+      "step": 207240
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425289807607355e-05,
+      "loss": 0.786,
+      "step": 207250
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425213642664555e-05,
+      "loss": 0.9637,
+      "step": 207260
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.425137477721754e-05,
+      "loss": 0.8695,
+      "step": 207270
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.4250613127789546e-05,
+      "loss": 0.9321,
+      "step": 207280
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.4249851478361537e-05,
+      "loss": 0.8764,
+      "step": 207290
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424908982893354e-05,
+      "loss": 0.9061,
+      "step": 207300
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424832817950554e-05,
+      "loss": 0.8628,
+      "step": 207310
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.4247566530077536e-05,
+      "loss": 0.9184,
+      "step": 207320
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424680488064953e-05,
+      "loss": 0.9183,
+      "step": 207330
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424604323122154e-05,
+      "loss": 0.8419,
+      "step": 207340
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424528158179353e-05,
+      "loss": 0.8945,
+      "step": 207350
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424451993236553e-05,
+      "loss": 0.874,
+      "step": 207360
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424375828293753e-05,
+      "loss": 0.8668,
+      "step": 207370
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424299663350953e-05,
+      "loss": 0.86,
+      "step": 207380
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424223498408153e-05,
+      "loss": 0.8616,
+      "step": 207390
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.424147333465353e-05,
+      "loss": 0.8662,
+      "step": 207400
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.4240711685225527e-05,
+      "loss": 0.8857,
+      "step": 207410
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.4239950035797524e-05,
+      "loss": 0.7913,
+      "step": 207420
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423918838636953e-05,
+      "loss": 0.8688,
+      "step": 207430
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423842673694152e-05,
+      "loss": 0.8391,
+      "step": 207440
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423766508751352e-05,
+      "loss": 0.9138,
+      "step": 207450
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423690343808552e-05,
+      "loss": 0.826,
+      "step": 207460
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423614178865752e-05,
+      "loss": 0.8836,
+      "step": 207470
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.4235380139229516e-05,
+      "loss": 0.8159,
+      "step": 207480
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423461848980152e-05,
+      "loss": 0.9175,
+      "step": 207490
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.423385684037351e-05,
+      "loss": 1.049,
+      "step": 207500
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4233095190945515e-05,
+      "loss": 0.9435,
+      "step": 207510
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.423233354151751e-05,
+      "loss": 0.8766,
+      "step": 207520
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.423157189208951e-05,
+      "loss": 0.9054,
+      "step": 207530
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.423081024266151e-05,
+      "loss": 0.864,
+      "step": 207540
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.423004859323351e-05,
+      "loss": 0.9548,
+      "step": 207550
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42292869438055e-05,
+      "loss": 1.0242,
+      "step": 207560
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4228525294377506e-05,
+      "loss": 0.9826,
+      "step": 207570
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4227763644949504e-05,
+      "loss": 0.9431,
+      "step": 207580
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42270019955215e-05,
+      "loss": 0.8933,
+      "step": 207590
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4226240346093505e-05,
+      "loss": 0.8636,
+      "step": 207600
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42254786966655e-05,
+      "loss": 0.8485,
+      "step": 207610
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42247170472375e-05,
+      "loss": 0.8687,
+      "step": 207620
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42239553978095e-05,
+      "loss": 0.785,
+      "step": 207630
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42231937483815e-05,
+      "loss": 0.9762,
+      "step": 207640
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.422243209895349e-05,
+      "loss": 0.9472,
+      "step": 207650
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.42216704495255e-05,
+      "loss": 0.8258,
+      "step": 207660
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4220908800097495e-05,
+      "loss": 0.8525,
+      "step": 207670
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.422014715066949e-05,
+      "loss": 0.8738,
+      "step": 207680
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.421938550124149e-05,
+      "loss": 0.9513,
+      "step": 207690
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4218623851813494e-05,
+      "loss": 0.7639,
+      "step": 207700
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.4217862202385484e-05,
+      "loss": 0.8798,
+      "step": 207710
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421710055295749e-05,
+      "loss": 0.8228,
+      "step": 207720
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421633890352948e-05,
+      "loss": 0.8934,
+      "step": 207730
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.4215577254101484e-05,
+      "loss": 0.918,
+      "step": 207740
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421481560467348e-05,
+      "loss": 0.9842,
+      "step": 207750
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421405395524548e-05,
+      "loss": 0.9699,
+      "step": 207760
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.4213292305817476e-05,
+      "loss": 0.8841,
+      "step": 207770
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421253065638948e-05,
+      "loss": 0.7756,
+      "step": 207780
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421176900696148e-05,
+      "loss": 0.8312,
+      "step": 207790
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.4211007357533475e-05,
+      "loss": 1.0015,
+      "step": 207800
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.421024570810548e-05,
+      "loss": 0.959,
+      "step": 207810
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420948405867747e-05,
+      "loss": 0.9029,
+      "step": 207820
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.4208722409249474e-05,
+      "loss": 0.8919,
+      "step": 207830
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420796075982147e-05,
+      "loss": 0.8697,
+      "step": 207840
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420719911039347e-05,
+      "loss": 0.8323,
+      "step": 207850
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420643746096547e-05,
+      "loss": 0.8767,
+      "step": 207860
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420567581153747e-05,
+      "loss": 0.8747,
+      "step": 207870
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420491416210946e-05,
+      "loss": 0.8726,
+      "step": 207880
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.4204152512681466e-05,
+      "loss": 0.8156,
+      "step": 207890
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420339086325346e-05,
+      "loss": 0.8486,
+      "step": 207900
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420262921382546e-05,
+      "loss": 0.8541,
+      "step": 207910
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420186756439746e-05,
+      "loss": 0.8643,
+      "step": 207920
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.420110591496946e-05,
+      "loss": 0.9012,
+      "step": 207930
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.420034426554145e-05,
+      "loss": 0.8796,
+      "step": 207940
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419958261611346e-05,
+      "loss": 0.8693,
+      "step": 207950
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4198820966685455e-05,
+      "loss": 1.0625,
+      "step": 207960
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419805931725745e-05,
+      "loss": 0.8466,
+      "step": 207970
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419729766782945e-05,
+      "loss": 0.7771,
+      "step": 207980
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4196536018401454e-05,
+      "loss": 0.894,
+      "step": 207990
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419577436897345e-05,
+      "loss": 0.8787,
+      "step": 208000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419501271954545e-05,
+      "loss": 0.8552,
+      "step": 208010
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419425107011745e-05,
+      "loss": 0.7808,
+      "step": 208020
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4193489420689444e-05,
+      "loss": 0.9287,
+      "step": 208030
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419272777126145e-05,
+      "loss": 0.8398,
+      "step": 208040
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4191966121833446e-05,
+      "loss": 0.858,
+      "step": 208050
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419120447240544e-05,
+      "loss": 0.7958,
+      "step": 208060
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.419044282297744e-05,
+      "loss": 0.7973,
+      "step": 208070
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4189681173549445e-05,
+      "loss": 0.9173,
+      "step": 208080
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4188919524121436e-05,
+      "loss": 0.8244,
+      "step": 208090
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.418815787469344e-05,
+      "loss": 0.8422,
+      "step": 208100
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.418739622526544e-05,
+      "loss": 0.9163,
+      "step": 208110
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4186634575837435e-05,
+      "loss": 0.9485,
+      "step": 208120
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.418587292640943e-05,
+      "loss": 0.8326,
+      "step": 208130
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.4185111276981436e-05,
+      "loss": 0.9053,
+      "step": 208140
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.418434962755343e-05,
+      "loss": 0.7397,
+      "step": 208150
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.418358797812543e-05,
+      "loss": 0.8536,
+      "step": 208160
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.418282632869743e-05,
+      "loss": 0.9681,
+      "step": 208170
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4182064679269426e-05,
+      "loss": 0.8919,
+      "step": 208180
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.418130302984143e-05,
+      "loss": 0.9608,
+      "step": 208190
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.418054138041342e-05,
+      "loss": 0.8999,
+      "step": 208200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4179779730985425e-05,
+      "loss": 0.9128,
+      "step": 208210
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417901808155742e-05,
+      "loss": 0.9579,
+      "step": 208220
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417825643212942e-05,
+      "loss": 0.9113,
+      "step": 208230
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417749478270142e-05,
+      "loss": 0.8155,
+      "step": 208240
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417673313327342e-05,
+      "loss": 0.8717,
+      "step": 208250
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417597148384541e-05,
+      "loss": 0.8346,
+      "step": 208260
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417520983441742e-05,
+      "loss": 0.8804,
+      "step": 208270
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4174448184989415e-05,
+      "loss": 0.9696,
+      "step": 208280
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417368653556141e-05,
+      "loss": 0.8225,
+      "step": 208290
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417292488613341e-05,
+      "loss": 0.8911,
+      "step": 208300
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4172163236705414e-05,
+      "loss": 0.8785,
+      "step": 208310
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4171401587277404e-05,
+      "loss": 0.8312,
+      "step": 208320
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.417063993784941e-05,
+      "loss": 0.8404,
+      "step": 208330
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4169878288421406e-05,
+      "loss": 0.8219,
+      "step": 208340
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4169116638993404e-05,
+      "loss": 0.8383,
+      "step": 208350
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.41683549895654e-05,
+      "loss": 0.9025,
+      "step": 208360
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.4167593340137405e-05,
+      "loss": 0.8135,
+      "step": 208370
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.41668316907094e-05,
+      "loss": 0.9792,
+      "step": 208380
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.41660700412814e-05,
+      "loss": 0.8562,
+      "step": 208390
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.4165308391853404e-05,
+      "loss": 0.8072,
+      "step": 208400
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.4164546742425395e-05,
+      "loss": 0.8191,
+      "step": 208410
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.41637850929974e-05,
+      "loss": 0.8596,
+      "step": 208420
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.41630234435694e-05,
+      "loss": 0.9467,
+      "step": 208430
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.4162261794141394e-05,
+      "loss": 0.9081,
+      "step": 208440
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.416150014471339e-05,
+      "loss": 0.8534,
+      "step": 208450
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.4160738495285396e-05,
+      "loss": 0.8656,
+      "step": 208460
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415997684585739e-05,
+      "loss": 0.8596,
+      "step": 208470
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415921519642939e-05,
+      "loss": 0.8952,
+      "step": 208480
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415845354700139e-05,
+      "loss": 0.9706,
+      "step": 208490
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.4157691897573386e-05,
+      "loss": 0.9455,
+      "step": 208500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415693024814538e-05,
+      "loss": 0.9049,
+      "step": 208510
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415616859871739e-05,
+      "loss": 0.9253,
+      "step": 208520
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415540694928938e-05,
+      "loss": 0.8537,
+      "step": 208530
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415464529986138e-05,
+      "loss": 0.8807,
+      "step": 208540
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415388365043338e-05,
+      "loss": 0.9933,
+      "step": 208550
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415312200100538e-05,
+      "loss": 0.8137,
+      "step": 208560
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.4152360351577375e-05,
+      "loss": 0.8594,
+      "step": 208570
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415159870214938e-05,
+      "loss": 0.7039,
+      "step": 208580
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.415083705272138e-05,
+      "loss": 0.7981,
+      "step": 208590
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4150075403293374e-05,
+      "loss": 0.852,
+      "step": 208600
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414931375386538e-05,
+      "loss": 0.9108,
+      "step": 208610
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414855210443737e-05,
+      "loss": 0.9119,
+      "step": 208620
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414779045500937e-05,
+      "loss": 0.9568,
+      "step": 208630
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414702880558137e-05,
+      "loss": 0.8254,
+      "step": 208640
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414626715615337e-05,
+      "loss": 0.7488,
+      "step": 208650
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4145505506725366e-05,
+      "loss": 0.8692,
+      "step": 208660
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414474385729736e-05,
+      "loss": 0.8812,
+      "step": 208670
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414398220786936e-05,
+      "loss": 0.9677,
+      "step": 208680
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4143220558441365e-05,
+      "loss": 0.9095,
+      "step": 208690
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4142458909013356e-05,
+      "loss": 0.8868,
+      "step": 208700
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414169725958536e-05,
+      "loss": 0.8854,
+      "step": 208710
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.414093561015736e-05,
+      "loss": 0.8114,
+      "step": 208720
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4140173960729355e-05,
+      "loss": 0.7933,
+      "step": 208730
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.413941231130135e-05,
+      "loss": 0.7529,
+      "step": 208740
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4138650661873356e-05,
+      "loss": 0.8298,
+      "step": 208750
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4137889012445354e-05,
+      "loss": 0.8794,
+      "step": 208760
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.413712736301735e-05,
+      "loss": 0.8672,
+      "step": 208770
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.413636571358935e-05,
+      "loss": 0.8681,
+      "step": 208780
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.4135604064161346e-05,
+      "loss": 0.7771,
+      "step": 208790
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.413484241473335e-05,
+      "loss": 0.9004,
+      "step": 208800
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.413408076530535e-05,
+      "loss": 0.8737,
+      "step": 208810
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.4133319115877345e-05,
+      "loss": 0.7547,
+      "step": 208820
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.413255746644934e-05,
+      "loss": 0.8916,
+      "step": 208830
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.413179581702135e-05,
+      "loss": 0.8802,
+      "step": 208840
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.413103416759334e-05,
+      "loss": 0.8945,
+      "step": 208850
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.413027251816534e-05,
+      "loss": 0.9515,
+      "step": 208860
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412951086873734e-05,
+      "loss": 0.8326,
+      "step": 208870
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412874921930934e-05,
+      "loss": 0.8304,
+      "step": 208880
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.4127987569881335e-05,
+      "loss": 0.8478,
+      "step": 208890
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412722592045334e-05,
+      "loss": 0.8113,
+      "step": 208900
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412646427102533e-05,
+      "loss": 0.7669,
+      "step": 208910
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.4125702621597334e-05,
+      "loss": 0.9092,
+      "step": 208920
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412494097216933e-05,
+      "loss": 0.8947,
+      "step": 208930
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412417932274133e-05,
+      "loss": 0.8402,
+      "step": 208940
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.4123417673313326e-05,
+      "loss": 0.9491,
+      "step": 208950
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412265602388533e-05,
+      "loss": 0.8354,
+      "step": 208960
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412189437445733e-05,
+      "loss": 0.8109,
+      "step": 208970
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.4121132725029325e-05,
+      "loss": 0.8304,
+      "step": 208980
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.412037107560133e-05,
+      "loss": 0.9114,
+      "step": 208990
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.411960942617332e-05,
+      "loss": 0.9092,
+      "step": 209000
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.4118847776745324e-05,
+      "loss": 0.8505,
+      "step": 209010
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.411808612731732e-05,
+      "loss": 0.8766,
+      "step": 209020
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.411732447788932e-05,
+      "loss": 0.8592,
+      "step": 209030
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.411656282846132e-05,
+      "loss": 0.8829,
+      "step": 209040
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.411580117903332e-05,
+      "loss": 0.7512,
+      "step": 209050
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.411503952960531e-05,
+      "loss": 0.8227,
+      "step": 209060
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.4114277880177316e-05,
+      "loss": 0.7673,
+      "step": 209070
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.4113516230749314e-05,
+      "loss": 0.8387,
+      "step": 209080
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.411275458132131e-05,
+      "loss": 0.8421,
+      "step": 209090
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.411199293189331e-05,
+      "loss": 0.9331,
+      "step": 209100
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.4111231282465306e-05,
+      "loss": 0.8804,
+      "step": 209110
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41104696330373e-05,
+      "loss": 0.8331,
+      "step": 209120
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.410970798360931e-05,
+      "loss": 0.8884,
+      "step": 209130
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.4108946334181305e-05,
+      "loss": 0.8403,
+      "step": 209140
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41081846847533e-05,
+      "loss": 0.8022,
+      "step": 209150
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41074230353253e-05,
+      "loss": 0.8803,
+      "step": 209160
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41066613858973e-05,
+      "loss": 0.8433,
+      "step": 209170
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41058997364693e-05,
+      "loss": 0.9139,
+      "step": 209180
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41051380870413e-05,
+      "loss": 0.7972,
+      "step": 209190
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41043764376133e-05,
+      "loss": 0.9264,
+      "step": 209200
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.4103614788185294e-05,
+      "loss": 0.8528,
+      "step": 209210
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.41028531387573e-05,
+      "loss": 0.9021,
+      "step": 209220
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.410209148932929e-05,
+      "loss": 0.9435,
+      "step": 209230
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.410132983990129e-05,
+      "loss": 0.809,
+      "step": 209240
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.410056819047329e-05,
+      "loss": 0.731,
+      "step": 209250
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409980654104529e-05,
+      "loss": 0.951,
+      "step": 209260
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4099044891617286e-05,
+      "loss": 0.8571,
+      "step": 209270
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409828324218929e-05,
+      "loss": 0.8906,
+      "step": 209280
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409752159276128e-05,
+      "loss": 0.8837,
+      "step": 209290
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4096759943333285e-05,
+      "loss": 0.8224,
+      "step": 209300
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409599829390528e-05,
+      "loss": 0.8398,
+      "step": 209310
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409523664447728e-05,
+      "loss": 0.9572,
+      "step": 209320
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409447499504928e-05,
+      "loss": 0.8752,
+      "step": 209330
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409371334562128e-05,
+      "loss": 0.7695,
+      "step": 209340
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409295169619328e-05,
+      "loss": 0.8138,
+      "step": 209350
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4092190046765276e-05,
+      "loss": 0.7885,
+      "step": 209360
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4091428397337274e-05,
+      "loss": 0.9292,
+      "step": 209370
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.409066674790927e-05,
+      "loss": 0.7834,
+      "step": 209380
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4089905098481276e-05,
+      "loss": 0.8816,
+      "step": 209390
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408914344905327e-05,
+      "loss": 0.9086,
+      "step": 209400
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408838179962527e-05,
+      "loss": 0.8878,
+      "step": 209410
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408762015019727e-05,
+      "loss": 1.1171,
+      "step": 209420
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408685850076927e-05,
+      "loss": 0.9559,
+      "step": 209430
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408609685134126e-05,
+      "loss": 0.8681,
+      "step": 209440
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408533520191327e-05,
+      "loss": 0.9045,
+      "step": 209450
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4084573552485265e-05,
+      "loss": 0.8349,
+      "step": 209460
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.408381190305726e-05,
+      "loss": 0.883,
+      "step": 209470
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.408305025362926e-05,
+      "loss": 0.9242,
+      "step": 209480
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4082288604201264e-05,
+      "loss": 0.8025,
+      "step": 209490
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4081526954773255e-05,
+      "loss": 0.8595,
+      "step": 209500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.408076530534526e-05,
+      "loss": 0.955,
+      "step": 209510
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4080003655917256e-05,
+      "loss": 0.9119,
+      "step": 209520
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4079242006489254e-05,
+      "loss": 0.8261,
+      "step": 209530
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407848035706125e-05,
+      "loss": 0.8542,
+      "step": 209540
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4077718707633255e-05,
+      "loss": 0.9085,
+      "step": 209550
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407695705820525e-05,
+      "loss": 0.7855,
+      "step": 209560
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407619540877725e-05,
+      "loss": 0.7791,
+      "step": 209570
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407543375934925e-05,
+      "loss": 0.9191,
+      "step": 209580
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4074672109921245e-05,
+      "loss": 0.8574,
+      "step": 209590
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407391046049325e-05,
+      "loss": 0.9334,
+      "step": 209600
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407314881106524e-05,
+      "loss": 0.8188,
+      "step": 209610
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4072387161637244e-05,
+      "loss": 0.9059,
+      "step": 209620
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407162551220924e-05,
+      "loss": 0.8134,
+      "step": 209630
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407086386278124e-05,
+      "loss": 0.9624,
+      "step": 209640
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.407010221335324e-05,
+      "loss": 0.9379,
+      "step": 209650
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.406934056392524e-05,
+      "loss": 0.8369,
+      "step": 209660
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.406857891449723e-05,
+      "loss": 0.8659,
+      "step": 209670
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4067817265069236e-05,
+      "loss": 0.9067,
+      "step": 209680
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.4067055615641234e-05,
+      "loss": 0.8962,
+      "step": 209690
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406629396621323e-05,
+      "loss": 0.9366,
+      "step": 209700
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406553231678523e-05,
+      "loss": 0.8546,
+      "step": 209710
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406477066735723e-05,
+      "loss": 0.7662,
+      "step": 209720
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406400901792923e-05,
+      "loss": 0.8502,
+      "step": 209730
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406324736850123e-05,
+      "loss": 0.8343,
+      "step": 209740
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4062485719073225e-05,
+      "loss": 0.8463,
+      "step": 209750
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406172406964522e-05,
+      "loss": 0.8802,
+      "step": 209760
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.406096242021723e-05,
+      "loss": 0.8409,
+      "step": 209770
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4060200770789224e-05,
+      "loss": 0.8987,
+      "step": 209780
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405943912136122e-05,
+      "loss": 0.8107,
+      "step": 209790
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405867747193322e-05,
+      "loss": 0.8728,
+      "step": 209800
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4057915822505223e-05,
+      "loss": 0.9404,
+      "step": 209810
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4057154173077214e-05,
+      "loss": 0.8896,
+      "step": 209820
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405639252364922e-05,
+      "loss": 0.8473,
+      "step": 209830
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4055630874221216e-05,
+      "loss": 0.7806,
+      "step": 209840
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405486922479321e-05,
+      "loss": 0.8094,
+      "step": 209850
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405410757536521e-05,
+      "loss": 0.9546,
+      "step": 209860
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4053345925937215e-05,
+      "loss": 0.8258,
+      "step": 209870
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.4052584276509206e-05,
+      "loss": 0.9417,
+      "step": 209880
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405182262708121e-05,
+      "loss": 0.8229,
+      "step": 209890
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.405106097765321e-05,
+      "loss": 0.8093,
+      "step": 209900
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4050299328225205e-05,
+      "loss": 0.8505,
+      "step": 209910
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.40495376787972e-05,
+      "loss": 0.75,
+      "step": 209920
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4048776029369207e-05,
+      "loss": 0.9233,
+      "step": 209930
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4048014379941204e-05,
+      "loss": 0.8371,
+      "step": 209940
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.40472527305132e-05,
+      "loss": 0.8273,
+      "step": 209950
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.40464910810852e-05,
+      "loss": 0.8708,
+      "step": 209960
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4045729431657196e-05,
+      "loss": 0.8356,
+      "step": 209970
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.40449677822292e-05,
+      "loss": 0.9842,
+      "step": 209980
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.40442061328012e-05,
+      "loss": 0.9634,
+      "step": 209990
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4043444483373196e-05,
+      "loss": 0.8038,
+      "step": 210000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.404268283394519e-05,
+      "loss": 0.8595,
+      "step": 210010
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.40419211845172e-05,
+      "loss": 0.8504,
+      "step": 210020
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.404115953508919e-05,
+      "loss": 0.8422,
+      "step": 210030
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.404039788566119e-05,
+      "loss": 0.9018,
+      "step": 210040
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.403963623623318e-05,
+      "loss": 0.9084,
+      "step": 210050
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.403887458680519e-05,
+      "loss": 0.8694,
+      "step": 210060
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4038112937377185e-05,
+      "loss": 0.8527,
+      "step": 210070
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.403735128794918e-05,
+      "loss": 1.0153,
+      "step": 210080
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.403658963852118e-05,
+      "loss": 0.9434,
+      "step": 210090
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.4035827989093184e-05,
+      "loss": 0.9288,
+      "step": 210100
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.403506633966518e-05,
+      "loss": 0.8493,
+      "step": 210110
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.403430469023718e-05,
+      "loss": 0.8921,
+      "step": 210120
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4033543040809176e-05,
+      "loss": 0.8356,
+      "step": 210130
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4032781391381174e-05,
+      "loss": 0.8312,
+      "step": 210140
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.403201974195318e-05,
+      "loss": 0.7726,
+      "step": 210150
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4031258092525175e-05,
+      "loss": 0.8648,
+      "step": 210160
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.403049644309717e-05,
+      "loss": 0.9477,
+      "step": 210170
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402973479366917e-05,
+      "loss": 0.8931,
+      "step": 210180
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4028973144241175e-05,
+      "loss": 0.8605,
+      "step": 210190
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4028211494813165e-05,
+      "loss": 0.8678,
+      "step": 210200
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402744984538517e-05,
+      "loss": 0.755,
+      "step": 210210
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402668819595717e-05,
+      "loss": 0.7829,
+      "step": 210220
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4025926546529164e-05,
+      "loss": 0.8483,
+      "step": 210230
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402516489710116e-05,
+      "loss": 0.907,
+      "step": 210240
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4024403247673166e-05,
+      "loss": 0.8095,
+      "step": 210250
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402364159824516e-05,
+      "loss": 0.8668,
+      "step": 210260
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402287994881716e-05,
+      "loss": 0.7863,
+      "step": 210270
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.402211829938916e-05,
+      "loss": 0.8554,
+      "step": 210280
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4021356649961156e-05,
+      "loss": 0.7935,
+      "step": 210290
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4020595000533154e-05,
+      "loss": 0.9296,
+      "step": 210300
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.401983335110516e-05,
+      "loss": 0.9412,
+      "step": 210310
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.4019071701677155e-05,
+      "loss": 0.8388,
+      "step": 210320
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.401831005224915e-05,
+      "loss": 0.892,
+      "step": 210330
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.401754840282115e-05,
+      "loss": 0.787,
+      "step": 210340
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401678675339315e-05,
+      "loss": 0.8389,
+      "step": 210350
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401602510396515e-05,
+      "loss": 0.857,
+      "step": 210360
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401526345453715e-05,
+      "loss": 0.8171,
+      "step": 210370
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401450180510915e-05,
+      "loss": 0.8355,
+      "step": 210380
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.4013740155681144e-05,
+      "loss": 0.7774,
+      "step": 210390
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401297850625315e-05,
+      "loss": 0.8465,
+      "step": 210400
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401221685682514e-05,
+      "loss": 0.8908,
+      "step": 210410
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.4011455207397143e-05,
+      "loss": 0.8998,
+      "step": 210420
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.401069355796914e-05,
+      "loss": 0.8182,
+      "step": 210430
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400993190854114e-05,
+      "loss": 0.809,
+      "step": 210440
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.4009170259113136e-05,
+      "loss": 0.9084,
+      "step": 210450
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400840860968514e-05,
+      "loss": 0.8881,
+      "step": 210460
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400764696025713e-05,
+      "loss": 0.9312,
+      "step": 210470
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.4006885310829135e-05,
+      "loss": 0.8031,
+      "step": 210480
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400612366140113e-05,
+      "loss": 0.9006,
+      "step": 210490
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400536201197313e-05,
+      "loss": 0.7692,
+      "step": 210500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400460036254513e-05,
+      "loss": 0.9213,
+      "step": 210510
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.4003838713117125e-05,
+      "loss": 0.8622,
+      "step": 210520
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400307706368913e-05,
+      "loss": 0.8734,
+      "step": 210530
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400231541426113e-05,
+      "loss": 0.9244,
+      "step": 210540
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.4001553764833124e-05,
+      "loss": 0.7901,
+      "step": 210550
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.400079211540512e-05,
+      "loss": 0.9019,
+      "step": 210560
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.4000030465977126e-05,
+      "loss": 0.8582,
+      "step": 210570
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.3999268816549116e-05,
+      "loss": 0.9223,
+      "step": 210580
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399850716712112e-05,
+      "loss": 0.8541,
+      "step": 210590
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399774551769312e-05,
+      "loss": 0.7655,
+      "step": 210600
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.3996983868265116e-05,
+      "loss": 0.9211,
+      "step": 210610
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399622221883711e-05,
+      "loss": 0.9087,
+      "step": 210620
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399546056940912e-05,
+      "loss": 0.9284,
+      "step": 210630
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399469891998111e-05,
+      "loss": 0.94,
+      "step": 210640
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399393727055311e-05,
+      "loss": 0.8685,
+      "step": 210650
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399317562112511e-05,
+      "loss": 0.803,
+      "step": 210660
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399241397169711e-05,
+      "loss": 0.8872,
+      "step": 210670
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.3991652322269105e-05,
+      "loss": 0.9334,
+      "step": 210680
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.399089067284111e-05,
+      "loss": 1.0292,
+      "step": 210690
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.3990129023413106e-05,
+      "loss": 0.8317,
+      "step": 210700
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.3989367373985104e-05,
+      "loss": 0.8195,
+      "step": 210710
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.39886057245571e-05,
+      "loss": 0.7951,
+      "step": 210720
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.39878440751291e-05,
+      "loss": 0.89,
+      "step": 210730
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.39870824257011e-05,
+      "loss": 0.8761,
+      "step": 210740
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.39863207762731e-05,
+      "loss": 0.864,
+      "step": 210750
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.39855591268451e-05,
+      "loss": 0.8997,
+      "step": 210760
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.3984797477417095e-05,
+      "loss": 0.9443,
+      "step": 210770
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.39840358279891e-05,
+      "loss": 1.0019,
+      "step": 210780
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.398327417856109e-05,
+      "loss": 0.8034,
+      "step": 210790
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.3982512529133095e-05,
+      "loss": 0.8008,
+      "step": 210800
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.398175087970509e-05,
+      "loss": 0.851,
+      "step": 210810
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.398098923027709e-05,
+      "loss": 0.8472,
+      "step": 210820
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.398022758084909e-05,
+      "loss": 0.7814,
+      "step": 210830
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397946593142109e-05,
+      "loss": 0.8711,
+      "step": 210840
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397870428199308e-05,
+      "loss": 0.8804,
+      "step": 210850
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.3977942632565086e-05,
+      "loss": 0.9142,
+      "step": 210860
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.3977180983137084e-05,
+      "loss": 0.8584,
+      "step": 210870
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397641933370908e-05,
+      "loss": 0.8196,
+      "step": 210880
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397565768428108e-05,
+      "loss": 0.8308,
+      "step": 210890
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397489603485308e-05,
+      "loss": 0.996,
+      "step": 210900
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397413438542508e-05,
+      "loss": 0.8603,
+      "step": 210910
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397337273599708e-05,
+      "loss": 0.7895,
+      "step": 210920
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.3972611086569075e-05,
+      "loss": 0.9014,
+      "step": 210930
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397184943714107e-05,
+      "loss": 1.0142,
+      "step": 210940
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397108778771308e-05,
+      "loss": 0.9369,
+      "step": 210950
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.397032613828507e-05,
+      "loss": 0.8282,
+      "step": 210960
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.396956448885707e-05,
+      "loss": 0.8507,
+      "step": 210970
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.396880283942907e-05,
+      "loss": 0.9748,
+      "step": 210980
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.396804119000107e-05,
+      "loss": 0.9022,
+      "step": 210990
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.3967279540573064e-05,
+      "loss": 0.8796,
+      "step": 211000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.396651789114507e-05,
+      "loss": 0.8605,
+      "step": 211010
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.396575624171706e-05,
+      "loss": 0.87,
+      "step": 211020
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.3964994592289063e-05,
+      "loss": 0.8684,
+      "step": 211030
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.396423294286106e-05,
+      "loss": 0.7234,
+      "step": 211040
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.396347129343306e-05,
+      "loss": 0.935,
+      "step": 211050
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.3962709644005056e-05,
+      "loss": 0.9496,
+      "step": 211060
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.396194799457706e-05,
+      "loss": 0.8199,
+      "step": 211070
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.396118634514905e-05,
+      "loss": 0.984,
+      "step": 211080
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.3960424695721055e-05,
+      "loss": 0.8408,
+      "step": 211090
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395966304629305e-05,
+      "loss": 0.9289,
+      "step": 211100
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395890139686505e-05,
+      "loss": 0.8881,
+      "step": 211110
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.3958139747437054e-05,
+      "loss": 0.8504,
+      "step": 211120
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395737809800905e-05,
+      "loss": 0.7846,
+      "step": 211130
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395661644858105e-05,
+      "loss": 0.7715,
+      "step": 211140
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395585479915305e-05,
+      "loss": 0.7979,
+      "step": 211150
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395509314972505e-05,
+      "loss": 0.8394,
+      "step": 211160
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395433150029704e-05,
+      "loss": 0.8923,
+      "step": 211170
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.3953569850869046e-05,
+      "loss": 0.9576,
+      "step": 211180
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395280820144104e-05,
+      "loss": 0.8716,
+      "step": 211190
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395204655201304e-05,
+      "loss": 0.8506,
+      "step": 211200
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395128490258504e-05,
+      "loss": 0.8752,
+      "step": 211210
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.395052325315704e-05,
+      "loss": 0.8628,
+      "step": 211220
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394976160372903e-05,
+      "loss": 0.8006,
+      "step": 211230
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394899995430104e-05,
+      "loss": 0.848,
+      "step": 211240
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3948238304873035e-05,
+      "loss": 0.8944,
+      "step": 211250
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394747665544503e-05,
+      "loss": 0.8147,
+      "step": 211260
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394671500601703e-05,
+      "loss": 0.897,
+      "step": 211270
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3945953356589034e-05,
+      "loss": 0.9197,
+      "step": 211280
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394519170716103e-05,
+      "loss": 0.8154,
+      "step": 211290
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394443005773303e-05,
+      "loss": 0.892,
+      "step": 211300
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3943668408305026e-05,
+      "loss": 0.8765,
+      "step": 211310
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3942906758877024e-05,
+      "loss": 0.9546,
+      "step": 211320
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394214510944903e-05,
+      "loss": 0.8633,
+      "step": 211330
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3941383460021026e-05,
+      "loss": 0.8412,
+      "step": 211340
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.394062181059302e-05,
+      "loss": 0.9041,
+      "step": 211350
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.393986016116502e-05,
+      "loss": 0.945,
+      "step": 211360
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3939098511737025e-05,
+      "loss": 0.7839,
+      "step": 211370
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3938336862309015e-05,
+      "loss": 0.9335,
+      "step": 211380
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.393757521288102e-05,
+      "loss": 0.8552,
+      "step": 211390
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.393681356345302e-05,
+      "loss": 0.7651,
+      "step": 211400
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3936051914025015e-05,
+      "loss": 0.855,
+      "step": 211410
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.393529026459701e-05,
+      "loss": 0.8164,
+      "step": 211420
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.393452861516901e-05,
+      "loss": 0.9462,
+      "step": 211430
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.393376696574101e-05,
+      "loss": 0.7776,
+      "step": 211440
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.393300531631301e-05,
+      "loss": 0.8501,
+      "step": 211450
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3932243666885e-05,
+      "loss": 0.922,
+      "step": 211460
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3931482017457006e-05,
+      "loss": 0.8401,
+      "step": 211470
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3930720368029004e-05,
+      "loss": 0.8872,
+      "step": 211480
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3929958718601e-05,
+      "loss": 0.8509,
+      "step": 211490
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3929197069173005e-05,
+      "loss": 0.8976,
+      "step": 211500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3928435419745e-05,
+      "loss": 0.973,
+      "step": 211510
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3927673770317e-05,
+      "loss": 0.8461,
+      "step": 211520
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3926912120889e-05,
+      "loss": 0.791,
+      "step": 211530
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3926150471461e-05,
+      "loss": 0.92,
+      "step": 211540
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.392538882203299e-05,
+      "loss": 0.8105,
+      "step": 211550
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3924627172605e-05,
+      "loss": 0.9093,
+      "step": 211560
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3923865523176994e-05,
+      "loss": 0.9363,
+      "step": 211570
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.392310387374899e-05,
+      "loss": 0.8717,
+      "step": 211580
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.392234222432099e-05,
+      "loss": 0.7699,
+      "step": 211590
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3921580574892994e-05,
+      "loss": 0.7944,
+      "step": 211600
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3920818925464984e-05,
+      "loss": 1.0156,
+      "step": 211610
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.392005727603699e-05,
+      "loss": 0.8509,
+      "step": 211620
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3919295626608986e-05,
+      "loss": 0.8352,
+      "step": 211630
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3918533977180983e-05,
+      "loss": 0.9646,
+      "step": 211640
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.391777232775298e-05,
+      "loss": 0.8708,
+      "step": 211650
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.3917010678324985e-05,
+      "loss": 0.814,
+      "step": 211660
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3916249028896976e-05,
+      "loss": 0.9252,
+      "step": 211670
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.391548737946898e-05,
+      "loss": 0.8347,
+      "step": 211680
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.391472573004098e-05,
+      "loss": 0.8727,
+      "step": 211690
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3913964080612975e-05,
+      "loss": 0.906,
+      "step": 211700
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.391320243118498e-05,
+      "loss": 0.8966,
+      "step": 211710
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.391244078175698e-05,
+      "loss": 0.8479,
+      "step": 211720
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3911679132328974e-05,
+      "loss": 0.9266,
+      "step": 211730
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.391091748290097e-05,
+      "loss": 0.7898,
+      "step": 211740
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3910155833472976e-05,
+      "loss": 0.8466,
+      "step": 211750
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390939418404497e-05,
+      "loss": 0.895,
+      "step": 211760
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390863253461697e-05,
+      "loss": 0.7314,
+      "step": 211770
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390787088518897e-05,
+      "loss": 0.836,
+      "step": 211780
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3907109235760966e-05,
+      "loss": 0.8544,
+      "step": 211790
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390634758633296e-05,
+      "loss": 0.8158,
+      "step": 211800
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390558593690497e-05,
+      "loss": 0.9193,
+      "step": 211810
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390482428747696e-05,
+      "loss": 0.8019,
+      "step": 211820
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390406263804896e-05,
+      "loss": 0.7862,
+      "step": 211830
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390330098862096e-05,
+      "loss": 0.8767,
+      "step": 211840
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390253933919296e-05,
+      "loss": 0.8305,
+      "step": 211850
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3901777689764955e-05,
+      "loss": 0.9299,
+      "step": 211860
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390101604033696e-05,
+      "loss": 0.8983,
+      "step": 211870
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.390025439090895e-05,
+      "loss": 0.8827,
+      "step": 211880
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3899492741480954e-05,
+      "loss": 0.9263,
+      "step": 211890
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389873109205295e-05,
+      "loss": 0.8307,
+      "step": 211900
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389796944262495e-05,
+      "loss": 0.9373,
+      "step": 211910
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389720779319695e-05,
+      "loss": 0.9333,
+      "step": 211920
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3896446143768944e-05,
+      "loss": 0.8367,
+      "step": 211930
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389568449434095e-05,
+      "loss": 0.8166,
+      "step": 211940
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3894922844912946e-05,
+      "loss": 0.8429,
+      "step": 211950
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389416119548494e-05,
+      "loss": 0.967,
+      "step": 211960
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389339954605694e-05,
+      "loss": 0.822,
+      "step": 211970
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3892637896628945e-05,
+      "loss": 0.7899,
+      "step": 211980
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3891876247200935e-05,
+      "loss": 0.8999,
+      "step": 211990
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389111459777294e-05,
+      "loss": 0.8803,
+      "step": 212000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.389035294834494e-05,
+      "loss": 0.8559,
+      "step": 212010
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3889591298916935e-05,
+      "loss": 1.0136,
+      "step": 212020
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.388882964948893e-05,
+      "loss": 0.9634,
+      "step": 212030
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3888068000060936e-05,
+      "loss": 0.9485,
+      "step": 212040
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.388730635063293e-05,
+      "loss": 0.942,
+      "step": 212050
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.388654470120493e-05,
+      "loss": 0.8549,
+      "step": 212060
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.388578305177693e-05,
+      "loss": 0.7964,
+      "step": 212070
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.3885021402348926e-05,
+      "loss": 0.8373,
+      "step": 212080
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.388425975292093e-05,
+      "loss": 0.965,
+      "step": 212090
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.388349810349293e-05,
+      "loss": 0.8319,
+      "step": 212100
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.3882736454064925e-05,
+      "loss": 0.9782,
+      "step": 212110
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.388197480463692e-05,
+      "loss": 0.8138,
+      "step": 212120
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.388121315520893e-05,
+      "loss": 0.8353,
+      "step": 212130
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.388045150578092e-05,
+      "loss": 0.7594,
+      "step": 212140
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387968985635292e-05,
+      "loss": 0.8662,
+      "step": 212150
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387892820692492e-05,
+      "loss": 0.9807,
+      "step": 212160
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387816655749692e-05,
+      "loss": 0.843,
+      "step": 212170
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.3877404908068914e-05,
+      "loss": 0.8329,
+      "step": 212180
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387664325864092e-05,
+      "loss": 0.8876,
+      "step": 212190
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387588160921291e-05,
+      "loss": 0.8255,
+      "step": 212200
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.3875119959784914e-05,
+      "loss": 0.83,
+      "step": 212210
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387435831035691e-05,
+      "loss": 0.9493,
+      "step": 212220
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387359666092891e-05,
+      "loss": 0.7953,
+      "step": 212230
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.3872835011500906e-05,
+      "loss": 0.9229,
+      "step": 212240
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.387207336207291e-05,
+      "loss": 0.8226,
+      "step": 212250
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.38713117126449e-05,
+      "loss": 0.8931,
+      "step": 212260
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.3870550063216905e-05,
+      "loss": 0.8078,
+      "step": 212270
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.38697884137889e-05,
+      "loss": 0.8493,
+      "step": 212280
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.38690267643609e-05,
+      "loss": 0.8136,
+      "step": 212290
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.3868265114932904e-05,
+      "loss": 0.8431,
+      "step": 212300
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.38675034655049e-05,
+      "loss": 0.8345,
+      "step": 212310
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.38667418160769e-05,
+      "loss": 0.8948,
+      "step": 212320
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.38659801666489e-05,
+      "loss": 0.8,
+      "step": 212330
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.3865218517220894e-05,
+      "loss": 0.9186,
+      "step": 212340
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.386445686779289e-05,
+      "loss": 0.9277,
+      "step": 212350
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.3863695218364896e-05,
+      "loss": 0.8358,
+      "step": 212360
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.386293356893689e-05,
+      "loss": 0.8633,
+      "step": 212370
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.386217191950889e-05,
+      "loss": 0.8929,
+      "step": 212380
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.386141027008089e-05,
+      "loss": 0.7702,
+      "step": 212390
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.3860648620652886e-05,
+      "loss": 0.8619,
+      "step": 212400
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385988697122488e-05,
+      "loss": 0.9417,
+      "step": 212410
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385912532179689e-05,
+      "loss": 1.03,
+      "step": 212420
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385836367236888e-05,
+      "loss": 0.8966,
+      "step": 212430
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385760202294088e-05,
+      "loss": 0.865,
+      "step": 212440
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385684037351288e-05,
+      "loss": 0.9576,
+      "step": 212450
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385607872408488e-05,
+      "loss": 0.8467,
+      "step": 212460
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.3855317074656875e-05,
+      "loss": 0.765,
+      "step": 212470
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385455542522888e-05,
+      "loss": 0.8946,
+      "step": 212480
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.3853793775800877e-05,
+      "loss": 0.9898,
+      "step": 212490
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.3853032126372874e-05,
+      "loss": 0.8733,
+      "step": 212500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385227047694488e-05,
+      "loss": 0.8804,
+      "step": 212510
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385150882751687e-05,
+      "loss": 0.9144,
+      "step": 212520
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.385074717808887e-05,
+      "loss": 0.8788,
+      "step": 212530
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384998552866087e-05,
+      "loss": 0.8348,
+      "step": 212540
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384922387923287e-05,
+      "loss": 1.006,
+      "step": 212550
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.3848462229804866e-05,
+      "loss": 0.9088,
+      "step": 212560
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384770058037687e-05,
+      "loss": 0.8814,
+      "step": 212570
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384693893094886e-05,
+      "loss": 0.9679,
+      "step": 212580
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.3846177281520865e-05,
+      "loss": 0.8335,
+      "step": 212590
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384541563209286e-05,
+      "loss": 0.8476,
+      "step": 212600
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384465398266486e-05,
+      "loss": 0.893,
+      "step": 212610
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384389233323686e-05,
+      "loss": 0.8741,
+      "step": 212620
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384313068380886e-05,
+      "loss": 0.8753,
+      "step": 212630
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384236903438085e-05,
+      "loss": 0.8352,
+      "step": 212640
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.3841607384952856e-05,
+      "loss": 0.7562,
+      "step": 212650
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.3840845735524854e-05,
+      "loss": 0.9076,
+      "step": 212660
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.384008408609685e-05,
+      "loss": 0.8909,
+      "step": 212670
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383932243666885e-05,
+      "loss": 0.9309,
+      "step": 212680
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383856078724085e-05,
+      "loss": 0.8787,
+      "step": 212690
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383779913781285e-05,
+      "loss": 0.9786,
+      "step": 212700
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383703748838485e-05,
+      "loss": 0.8727,
+      "step": 212710
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383627583895685e-05,
+      "loss": 0.9255,
+      "step": 212720
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383551418952884e-05,
+      "loss": 0.8711,
+      "step": 212730
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.383475254010085e-05,
+      "loss": 0.9524,
+      "step": 212740
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.3833990890672845e-05,
+      "loss": 0.7676,
+      "step": 212750
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.383322924124484e-05,
+      "loss": 0.911,
+      "step": 212760
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.383246759181684e-05,
+      "loss": 0.8963,
+      "step": 212770
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3831705942388844e-05,
+      "loss": 1.0218,
+      "step": 212780
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3830944292960834e-05,
+      "loss": 0.8196,
+      "step": 212790
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.383018264353284e-05,
+      "loss": 0.8756,
+      "step": 212800
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382942099410483e-05,
+      "loss": 0.9082,
+      "step": 212810
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3828659344676834e-05,
+      "loss": 0.962,
+      "step": 212820
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382789769524883e-05,
+      "loss": 0.9356,
+      "step": 212830
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382713604582083e-05,
+      "loss": 0.9011,
+      "step": 212840
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3826374396392826e-05,
+      "loss": 0.8666,
+      "step": 212850
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382561274696483e-05,
+      "loss": 0.7689,
+      "step": 212860
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382485109753683e-05,
+      "loss": 0.8739,
+      "step": 212870
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3824089448108825e-05,
+      "loss": 0.9475,
+      "step": 212880
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382332779868083e-05,
+      "loss": 0.8623,
+      "step": 212890
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382256614925282e-05,
+      "loss": 0.9567,
+      "step": 212900
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3821804499824824e-05,
+      "loss": 0.8454,
+      "step": 212910
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382104285039682e-05,
+      "loss": 0.8613,
+      "step": 212920
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.382028120096882e-05,
+      "loss": 0.8759,
+      "step": 212930
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.381951955154082e-05,
+      "loss": 0.8295,
+      "step": 212940
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.381875790211282e-05,
+      "loss": 0.8656,
+      "step": 212950
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.381799625268481e-05,
+      "loss": 0.8435,
+      "step": 212960
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.3817234603256816e-05,
+      "loss": 0.8986,
+      "step": 212970
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3816472953828813e-05,
+      "loss": 0.8502,
+      "step": 212980
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.381571130440081e-05,
+      "loss": 0.8555,
+      "step": 212990
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.381494965497281e-05,
+      "loss": 0.8804,
+      "step": 213000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.381418800554481e-05,
+      "loss": 0.9234,
+      "step": 213010
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38134263561168e-05,
+      "loss": 0.8429,
+      "step": 213020
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.381266470668881e-05,
+      "loss": 0.8793,
+      "step": 213030
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3811903057260805e-05,
+      "loss": 0.9285,
+      "step": 213040
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38111414078328e-05,
+      "loss": 0.8365,
+      "step": 213050
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38103797584048e-05,
+      "loss": 0.9147,
+      "step": 213060
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3809618108976804e-05,
+      "loss": 0.815,
+      "step": 213070
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38088564595488e-05,
+      "loss": 0.8495,
+      "step": 213080
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38080948101208e-05,
+      "loss": 0.8196,
+      "step": 213090
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38073331606928e-05,
+      "loss": 0.8573,
+      "step": 213100
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3806571511264794e-05,
+      "loss": 0.8677,
+      "step": 213110
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.38058098618368e-05,
+      "loss": 0.8049,
+      "step": 213120
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3805048212408796e-05,
+      "loss": 0.8574,
+      "step": 213130
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.380428656298079e-05,
+      "loss": 0.7869,
+      "step": 213140
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.380352491355279e-05,
+      "loss": 0.8121,
+      "step": 213150
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3802763264124795e-05,
+      "loss": 0.8436,
+      "step": 213160
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.3802001614696786e-05,
+      "loss": 0.8811,
+      "step": 213170
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.380123996526879e-05,
+      "loss": 0.881,
+      "step": 213180
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.380047831584079e-05,
+      "loss": 0.7987,
+      "step": 213190
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3799716666412785e-05,
+      "loss": 0.8644,
+      "step": 213200
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379895501698478e-05,
+      "loss": 0.9,
+      "step": 213210
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3798193367556787e-05,
+      "loss": 0.8741,
+      "step": 213220
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379743171812878e-05,
+      "loss": 0.819,
+      "step": 213230
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379667006870078e-05,
+      "loss": 0.8152,
+      "step": 213240
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379590841927278e-05,
+      "loss": 0.8876,
+      "step": 213250
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3795146769844776e-05,
+      "loss": 0.8765,
+      "step": 213260
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3794385120416774e-05,
+      "loss": 0.814,
+      "step": 213270
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379362347098877e-05,
+      "loss": 0.8625,
+      "step": 213280
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3792861821560776e-05,
+      "loss": 0.8482,
+      "step": 213290
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379210017213277e-05,
+      "loss": 0.7916,
+      "step": 213300
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379133852270477e-05,
+      "loss": 0.7903,
+      "step": 213310
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.379057687327677e-05,
+      "loss": 0.8137,
+      "step": 213320
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.378981522384877e-05,
+      "loss": 0.9879,
+      "step": 213330
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.378905357442076e-05,
+      "loss": 0.8612,
+      "step": 213340
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.378829192499277e-05,
+      "loss": 0.9132,
+      "step": 213350
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3787530275564765e-05,
+      "loss": 0.7954,
+      "step": 213360
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.378676862613676e-05,
+      "loss": 0.85,
+      "step": 213370
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.378600697670876e-05,
+      "loss": 0.8421,
+      "step": 213380
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3785245327280764e-05,
+      "loss": 0.8279,
+      "step": 213390
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3784483677852754e-05,
+      "loss": 0.8793,
+      "step": 213400
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.378372202842476e-05,
+      "loss": 0.9055,
+      "step": 213410
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3782960378996756e-05,
+      "loss": 0.8819,
+      "step": 213420
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3782198729568754e-05,
+      "loss": 0.8879,
+      "step": 213430
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.378143708014075e-05,
+      "loss": 0.9077,
+      "step": 213440
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3780675430712755e-05,
+      "loss": 0.8537,
+      "step": 213450
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377991378128475e-05,
+      "loss": 0.846,
+      "step": 213460
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377915213185675e-05,
+      "loss": 0.9235,
+      "step": 213470
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377839048242875e-05,
+      "loss": 0.8755,
+      "step": 213480
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3777628833000745e-05,
+      "loss": 0.8356,
+      "step": 213490
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377686718357275e-05,
+      "loss": 0.8089,
+      "step": 213500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377610553414475e-05,
+      "loss": 0.9181,
+      "step": 213510
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3775343884716744e-05,
+      "loss": 0.8341,
+      "step": 213520
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377458223528874e-05,
+      "loss": 1.0464,
+      "step": 213530
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3773820585860746e-05,
+      "loss": 0.8082,
+      "step": 213540
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377305893643274e-05,
+      "loss": 0.9237,
+      "step": 213550
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377229728700474e-05,
+      "loss": 0.8162,
+      "step": 213560
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.377153563757674e-05,
+      "loss": 0.8457,
+      "step": 213570
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3770773988148736e-05,
+      "loss": 0.9609,
+      "step": 213580
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3770012338720733e-05,
+      "loss": 0.8123,
+      "step": 213590
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.376925068929274e-05,
+      "loss": 0.8202,
+      "step": 213600
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.376848903986473e-05,
+      "loss": 0.9285,
+      "step": 213610
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.376772739043673e-05,
+      "loss": 0.9774,
+      "step": 213620
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.376696574100873e-05,
+      "loss": 0.8498,
+      "step": 213630
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.376620409158073e-05,
+      "loss": 1.0365,
+      "step": 213640
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3765442442152725e-05,
+      "loss": 0.8934,
+      "step": 213650
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.376468079272473e-05,
+      "loss": 0.792,
+      "step": 213660
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.376391914329673e-05,
+      "loss": 0.8847,
+      "step": 213670
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3763157493868724e-05,
+      "loss": 0.779,
+      "step": 213680
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.376239584444073e-05,
+      "loss": 0.858,
+      "step": 213690
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.376163419501272e-05,
+      "loss": 0.9464,
+      "step": 213700
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.376087254558472e-05,
+      "loss": 0.8711,
+      "step": 213710
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3760110896156714e-05,
+      "loss": 0.857,
+      "step": 213720
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.375934924672872e-05,
+      "loss": 0.8936,
+      "step": 213730
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3758587597300716e-05,
+      "loss": 1.0312,
+      "step": 213740
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.375782594787271e-05,
+      "loss": 0.8726,
+      "step": 213750
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.375706429844471e-05,
+      "loss": 0.9141,
+      "step": 213760
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3756302649016715e-05,
+      "loss": 0.9156,
+      "step": 213770
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3755540999588706e-05,
+      "loss": 0.8883,
+      "step": 213780
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.375477935016071e-05,
+      "loss": 0.7831,
+      "step": 213790
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.375401770073271e-05,
+      "loss": 0.7906,
+      "step": 213800
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3753256051304705e-05,
+      "loss": 0.8914,
+      "step": 213810
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.37524944018767e-05,
+      "loss": 0.8875,
+      "step": 213820
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3751732752448707e-05,
+      "loss": 0.8949,
+      "step": 213830
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.3750971103020704e-05,
+      "loss": 0.8788,
+      "step": 213840
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.37502094535927e-05,
+      "loss": 0.8171,
+      "step": 213850
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.37494478041647e-05,
+      "loss": 0.8324,
+      "step": 213860
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.3748686154736696e-05,
+      "loss": 0.797,
+      "step": 213870
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.37479245053087e-05,
+      "loss": 0.8397,
+      "step": 213880
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.37471628558807e-05,
+      "loss": 0.8335,
+      "step": 213890
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.3746401206452696e-05,
+      "loss": 0.8727,
+      "step": 213900
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.374563955702469e-05,
+      "loss": 0.8479,
+      "step": 213910
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.37448779075967e-05,
+      "loss": 0.8351,
+      "step": 213920
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.374411625816869e-05,
+      "loss": 0.8859,
+      "step": 213930
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.374335460874069e-05,
+      "loss": 0.9584,
+      "step": 213940
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.374259295931269e-05,
+      "loss": 0.9047,
+      "step": 213950
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.374183130988469e-05,
+      "loss": 0.8231,
+      "step": 213960
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.3741069660456685e-05,
+      "loss": 0.9539,
+      "step": 213970
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.374030801102869e-05,
+      "loss": 0.901,
+      "step": 213980
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.373954636160068e-05,
+      "loss": 0.8992,
+      "step": 213990
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.3738784712172684e-05,
+      "loss": 0.804,
+      "step": 214000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.373802306274468e-05,
+      "loss": 0.9098,
+      "step": 214010
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.373726141331668e-05,
+      "loss": 0.906,
+      "step": 214020
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.3736499763888676e-05,
+      "loss": 0.8245,
+      "step": 214030
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.373573811446068e-05,
+      "loss": 0.6983,
+      "step": 214040
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.373497646503268e-05,
+      "loss": 0.8129,
+      "step": 214050
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.3734214815604675e-05,
+      "loss": 0.7742,
+      "step": 214060
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.373345316617667e-05,
+      "loss": 0.871,
+      "step": 214070
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.373269151674867e-05,
+      "loss": 0.8835,
+      "step": 214080
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.3731929867320675e-05,
+      "loss": 0.8103,
+      "step": 214090
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.373116821789267e-05,
+      "loss": 0.8691,
+      "step": 214100
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.373040656846467e-05,
+      "loss": 0.9095,
+      "step": 214110
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372964491903667e-05,
+      "loss": 0.8877,
+      "step": 214120
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372888326960867e-05,
+      "loss": 0.8401,
+      "step": 214130
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372812162018066e-05,
+      "loss": 0.7898,
+      "step": 214140
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.3727359970752666e-05,
+      "loss": 0.8633,
+      "step": 214150
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.3726598321324664e-05,
+      "loss": 0.8669,
+      "step": 214160
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372583667189666e-05,
+      "loss": 0.7691,
+      "step": 214170
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372507502246866e-05,
+      "loss": 0.9013,
+      "step": 214180
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.3724313373040656e-05,
+      "loss": 0.8195,
+      "step": 214190
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.3723551723612653e-05,
+      "loss": 0.8547,
+      "step": 214200
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372279007418466e-05,
+      "loss": 0.8371,
+      "step": 214210
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.3722028424756655e-05,
+      "loss": 0.8871,
+      "step": 214220
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372126677532865e-05,
+      "loss": 0.9156,
+      "step": 214230
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.372050512590065e-05,
+      "loss": 0.7815,
+      "step": 214240
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.371974347647265e-05,
+      "loss": 0.9154,
+      "step": 214250
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.371898182704465e-05,
+      "loss": 0.9769,
+      "step": 214260
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.371822017761665e-05,
+      "loss": 0.8289,
+      "step": 214270
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.371745852818865e-05,
+      "loss": 0.832,
+      "step": 214280
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3716696878760644e-05,
+      "loss": 0.9585,
+      "step": 214290
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.371593522933265e-05,
+      "loss": 0.8551,
+      "step": 214300
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.371517357990464e-05,
+      "loss": 0.8128,
+      "step": 214310
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3714411930476643e-05,
+      "loss": 0.9364,
+      "step": 214320
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.371365028104864e-05,
+      "loss": 0.8897,
+      "step": 214330
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.371288863162064e-05,
+      "loss": 0.9035,
+      "step": 214340
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3712126982192636e-05,
+      "loss": 0.9126,
+      "step": 214350
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.371136533276464e-05,
+      "loss": 0.8389,
+      "step": 214360
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.371060368333663e-05,
+      "loss": 0.8686,
+      "step": 214370
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3709842033908635e-05,
+      "loss": 0.7713,
+      "step": 214380
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370908038448063e-05,
+      "loss": 0.8394,
+      "step": 214390
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370831873505263e-05,
+      "loss": 0.7706,
+      "step": 214400
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370755708562463e-05,
+      "loss": 0.8715,
+      "step": 214410
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370679543619663e-05,
+      "loss": 0.9124,
+      "step": 214420
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370603378676863e-05,
+      "loss": 0.9554,
+      "step": 214430
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3705272137340627e-05,
+      "loss": 0.8274,
+      "step": 214440
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3704510487912624e-05,
+      "loss": 0.899,
+      "step": 214450
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370374883848462e-05,
+      "loss": 0.8399,
+      "step": 214460
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3702987189056626e-05,
+      "loss": 0.7945,
+      "step": 214470
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370222553962862e-05,
+      "loss": 0.8574,
+      "step": 214480
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370146389020062e-05,
+      "loss": 0.798,
+      "step": 214490
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.370070224077262e-05,
+      "loss": 0.9196,
+      "step": 214500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.369994059134462e-05,
+      "loss": 0.9126,
+      "step": 214510
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.369917894191661e-05,
+      "loss": 0.7908,
+      "step": 214520
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.369841729248862e-05,
+      "loss": 0.8252,
+      "step": 214530
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3697655643060615e-05,
+      "loss": 0.9649,
+      "step": 214540
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.369689399363261e-05,
+      "loss": 0.8356,
+      "step": 214550
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.369613234420461e-05,
+      "loss": 0.878,
+      "step": 214560
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3695370694776614e-05,
+      "loss": 0.8224,
+      "step": 214570
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3694609045348605e-05,
+      "loss": 0.8936,
+      "step": 214580
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.369384739592061e-05,
+      "loss": 0.7911,
+      "step": 214590
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3693085746492606e-05,
+      "loss": 0.8665,
+      "step": 214600
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3692324097064604e-05,
+      "loss": 0.8357,
+      "step": 214610
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.36915624476366e-05,
+      "loss": 0.7698,
+      "step": 214620
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3690800798208606e-05,
+      "loss": 0.8135,
+      "step": 214630
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.36900391487806e-05,
+      "loss": 0.8406,
+      "step": 214640
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.36892774993526e-05,
+      "loss": 0.8009,
+      "step": 214650
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.36885158499246e-05,
+      "loss": 0.8679,
+      "step": 214660
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3687754200496595e-05,
+      "loss": 0.874,
+      "step": 214670
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.36869925510686e-05,
+      "loss": 0.9001,
+      "step": 214680
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.368623090164059e-05,
+      "loss": 0.7856,
+      "step": 214690
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.3685469252212595e-05,
+      "loss": 0.8976,
+      "step": 214700
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.368470760278459e-05,
+      "loss": 0.8391,
+      "step": 214710
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.368394595335659e-05,
+      "loss": 0.8856,
+      "step": 214720
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.368318430392859e-05,
+      "loss": 0.8461,
+      "step": 214730
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.368242265450059e-05,
+      "loss": 0.8605,
+      "step": 214740
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.368166100507258e-05,
+      "loss": 0.8622,
+      "step": 214750
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3680899355644586e-05,
+      "loss": 0.869,
+      "step": 214760
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3680137706216584e-05,
+      "loss": 0.8353,
+      "step": 214770
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367937605678858e-05,
+      "loss": 0.8909,
+      "step": 214780
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367861440736058e-05,
+      "loss": 0.9471,
+      "step": 214790
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367785275793258e-05,
+      "loss": 0.9465,
+      "step": 214800
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367709110850458e-05,
+      "loss": 0.8615,
+      "step": 214810
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367632945907658e-05,
+      "loss": 0.7898,
+      "step": 214820
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3675567809648575e-05,
+      "loss": 0.8424,
+      "step": 214830
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367480616022057e-05,
+      "loss": 0.8636,
+      "step": 214840
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367404451079258e-05,
+      "loss": 0.8204,
+      "step": 214850
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3673282861364574e-05,
+      "loss": 0.8751,
+      "step": 214860
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367252121193657e-05,
+      "loss": 0.9062,
+      "step": 214870
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.367175956250857e-05,
+      "loss": 0.7803,
+      "step": 214880
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3670997913080574e-05,
+      "loss": 0.807,
+      "step": 214890
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3670236263652564e-05,
+      "loss": 0.853,
+      "step": 214900
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.366947461422457e-05,
+      "loss": 0.9237,
+      "step": 214910
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3668712964796566e-05,
+      "loss": 0.8996,
+      "step": 214920
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.3667951315368563e-05,
+      "loss": 0.8674,
+      "step": 214930
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.366718966594056e-05,
+      "loss": 0.8155,
+      "step": 214940
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3666428016512565e-05,
+      "loss": 0.9134,
+      "step": 214950
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3665666367084556e-05,
+      "loss": 0.8201,
+      "step": 214960
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.366490471765656e-05,
+      "loss": 0.8021,
+      "step": 214970
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.366414306822856e-05,
+      "loss": 0.8555,
+      "step": 214980
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3663381418800555e-05,
+      "loss": 0.9325,
+      "step": 214990
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.366261976937255e-05,
+      "loss": 0.8586,
+      "step": 215000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.366185811994456e-05,
+      "loss": 0.8575,
+      "step": 215010
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3661096470516554e-05,
+      "loss": 0.8955,
+      "step": 215020
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.366033482108855e-05,
+      "loss": 1.0189,
+      "step": 215030
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365957317166055e-05,
+      "loss": 0.8969,
+      "step": 215040
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3658811522232547e-05,
+      "loss": 0.8746,
+      "step": 215050
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365804987280455e-05,
+      "loss": 0.7919,
+      "step": 215060
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365728822337655e-05,
+      "loss": 1.0477,
+      "step": 215070
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3656526573948546e-05,
+      "loss": 0.85,
+      "step": 215080
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365576492452054e-05,
+      "loss": 0.8966,
+      "step": 215090
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365500327509254e-05,
+      "loss": 0.8052,
+      "step": 215100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365424162566454e-05,
+      "loss": 0.7913,
+      "step": 215110
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365347997623654e-05,
+      "loss": 0.926,
+      "step": 215120
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365271832680853e-05,
+      "loss": 0.809,
+      "step": 215130
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365195667738054e-05,
+      "loss": 0.8818,
+      "step": 215140
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.3651195027952535e-05,
+      "loss": 0.9526,
+      "step": 215150
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.365043337852453e-05,
+      "loss": 0.864,
+      "step": 215160
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364967172909653e-05,
+      "loss": 0.8588,
+      "step": 215170
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3648910079668534e-05,
+      "loss": 0.9207,
+      "step": 215180
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364814843024053e-05,
+      "loss": 0.9728,
+      "step": 215190
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364738678081253e-05,
+      "loss": 0.7914,
+      "step": 215200
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3646625131384526e-05,
+      "loss": 0.8654,
+      "step": 215210
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3645863481956524e-05,
+      "loss": 0.8456,
+      "step": 215220
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364510183252853e-05,
+      "loss": 0.875,
+      "step": 215230
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3644340183100526e-05,
+      "loss": 0.8604,
+      "step": 215240
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364357853367252e-05,
+      "loss": 0.8135,
+      "step": 215250
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364281688424452e-05,
+      "loss": 0.9242,
+      "step": 215260
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3642055234816525e-05,
+      "loss": 0.9065,
+      "step": 215270
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3641293585388515e-05,
+      "loss": 0.9061,
+      "step": 215280
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.364053193596052e-05,
+      "loss": 0.8506,
+      "step": 215290
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.363977028653252e-05,
+      "loss": 0.8515,
+      "step": 215300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3639008637104515e-05,
+      "loss": 0.7982,
+      "step": 215310
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.363824698767651e-05,
+      "loss": 0.7888,
+      "step": 215320
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3637485338248516e-05,
+      "loss": 0.9234,
+      "step": 215330
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.363672368882051e-05,
+      "loss": 0.7744,
+      "step": 215340
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.363596203939251e-05,
+      "loss": 0.8315,
+      "step": 215350
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.363520038996451e-05,
+      "loss": 0.7495,
+      "step": 215360
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3634438740536506e-05,
+      "loss": 0.8586,
+      "step": 215370
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.3633677091108504e-05,
+      "loss": 0.8877,
+      "step": 215380
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.363291544168051e-05,
+      "loss": 0.7999,
+      "step": 215390
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3632153792252505e-05,
+      "loss": 0.8563,
+      "step": 215400
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36313921428245e-05,
+      "loss": 0.9971,
+      "step": 215410
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36306304933965e-05,
+      "loss": 0.8058,
+      "step": 215420
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36298688439685e-05,
+      "loss": 0.7656,
+      "step": 215430
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36291071945405e-05,
+      "loss": 0.9231,
+      "step": 215440
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36283455451125e-05,
+      "loss": 0.9401,
+      "step": 215450
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36275838956845e-05,
+      "loss": 0.9272,
+      "step": 215460
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3626822246256494e-05,
+      "loss": 0.9636,
+      "step": 215470
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.36260605968285e-05,
+      "loss": 0.7973,
+      "step": 215480
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.362529894740049e-05,
+      "loss": 0.9122,
+      "step": 215490
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3624537297972494e-05,
+      "loss": 0.8614,
+      "step": 215500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.362377564854449e-05,
+      "loss": 0.9297,
+      "step": 215510
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.362301399911649e-05,
+      "loss": 0.9064,
+      "step": 215520
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3622252349688486e-05,
+      "loss": 0.832,
+      "step": 215530
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.362149070026049e-05,
+      "loss": 0.8246,
+      "step": 215540
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.362072905083248e-05,
+      "loss": 0.871,
+      "step": 215550
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3619967401404485e-05,
+      "loss": 0.8779,
+      "step": 215560
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3619205751976476e-05,
+      "loss": 0.8307,
+      "step": 215570
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.361844410254848e-05,
+      "loss": 0.8665,
+      "step": 215580
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.361768245312048e-05,
+      "loss": 0.9152,
+      "step": 215590
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.3616920803692475e-05,
+      "loss": 0.8935,
+      "step": 215600
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.361615915426448e-05,
+      "loss": 0.9283,
+      "step": 215610
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.361539750483648e-05,
+      "loss": 0.7983,
+      "step": 215620
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3614635855408474e-05,
+      "loss": 0.8486,
+      "step": 215630
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.361387420598047e-05,
+      "loss": 0.8281,
+      "step": 215640
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3613112556552476e-05,
+      "loss": 0.7763,
+      "step": 215650
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3612350907124467e-05,
+      "loss": 0.7603,
+      "step": 215660
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.361158925769647e-05,
+      "loss": 0.8259,
+      "step": 215670
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.361082760826847e-05,
+      "loss": 0.9653,
+      "step": 215680
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3610065958840466e-05,
+      "loss": 0.8653,
+      "step": 215690
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360930430941246e-05,
+      "loss": 0.9081,
+      "step": 215700
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360854265998447e-05,
+      "loss": 0.7873,
+      "step": 215710
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360778101055646e-05,
+      "loss": 1.0195,
+      "step": 215720
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360701936112846e-05,
+      "loss": 0.8754,
+      "step": 215730
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360625771170046e-05,
+      "loss": 0.8658,
+      "step": 215740
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360549606227246e-05,
+      "loss": 0.9473,
+      "step": 215750
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3604734412844455e-05,
+      "loss": 1.0328,
+      "step": 215760
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360397276341646e-05,
+      "loss": 0.8301,
+      "step": 215770
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3603211113988456e-05,
+      "loss": 0.9344,
+      "step": 215780
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.3602449464560454e-05,
+      "loss": 0.913,
+      "step": 215790
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360168781513245e-05,
+      "loss": 0.7946,
+      "step": 215800
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360092616570445e-05,
+      "loss": 0.8611,
+      "step": 215810
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.360016451627645e-05,
+      "loss": 0.825,
+      "step": 215820
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359940286684845e-05,
+      "loss": 0.777,
+      "step": 215830
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359864121742045e-05,
+      "loss": 0.9037,
+      "step": 215840
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.3597879567992446e-05,
+      "loss": 0.8693,
+      "step": 215850
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359711791856445e-05,
+      "loss": 0.7327,
+      "step": 215860
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359635626913644e-05,
+      "loss": 0.8204,
+      "step": 215870
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.3595594619708445e-05,
+      "loss": 0.8507,
+      "step": 215880
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359483297028044e-05,
+      "loss": 0.9224,
+      "step": 215890
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359407132085244e-05,
+      "loss": 0.9171,
+      "step": 215900
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359330967142444e-05,
+      "loss": 0.8531,
+      "step": 215910
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359254802199644e-05,
+      "loss": 0.7749,
+      "step": 215920
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.359178637256843e-05,
+      "loss": 0.8189,
+      "step": 215930
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.3591024723140436e-05,
+      "loss": 0.869,
+      "step": 215940
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.3590263073712434e-05,
+      "loss": 0.9353,
+      "step": 215950
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358950142428443e-05,
+      "loss": 0.8854,
+      "step": 215960
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358873977485643e-05,
+      "loss": 0.8121,
+      "step": 215970
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358797812542843e-05,
+      "loss": 0.7952,
+      "step": 215980
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358721647600043e-05,
+      "loss": 0.8052,
+      "step": 215990
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358645482657243e-05,
+      "loss": 0.8508,
+      "step": 216000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.3585693177144425e-05,
+      "loss": 0.9303,
+      "step": 216010
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358493152771642e-05,
+      "loss": 0.8565,
+      "step": 216020
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358416987828843e-05,
+      "loss": 0.9653,
+      "step": 216030
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.358340822886042e-05,
+      "loss": 0.8105,
+      "step": 216040
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.358264657943242e-05,
+      "loss": 0.8666,
+      "step": 216050
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.358188493000442e-05,
+      "loss": 0.8718,
+      "step": 216060
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.358112328057642e-05,
+      "loss": 0.84,
+      "step": 216070
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.3580361631148414e-05,
+      "loss": 0.8016,
+      "step": 216080
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.357959998172042e-05,
+      "loss": 0.93,
+      "step": 216090
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.357883833229241e-05,
+      "loss": 0.8301,
+      "step": 216100
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.3578076682864414e-05,
+      "loss": 0.8659,
+      "step": 216110
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.357731503343641e-05,
+      "loss": 0.9137,
+      "step": 216120
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.357655338400841e-05,
+      "loss": 0.7827,
+      "step": 216130
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.3575791734580406e-05,
+      "loss": 0.9586,
+      "step": 216140
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.357503008515241e-05,
+      "loss": 0.899,
+      "step": 216150
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35742684357244e-05,
+      "loss": 0.9236,
+      "step": 216160
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.3573506786296405e-05,
+      "loss": 0.812,
+      "step": 216170
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35727451368684e-05,
+      "loss": 0.863,
+      "step": 216180
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35719834874404e-05,
+      "loss": 0.912,
+      "step": 216190
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.3571221838012404e-05,
+      "loss": 0.9318,
+      "step": 216200
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35704601885844e-05,
+      "loss": 0.8352,
+      "step": 216210
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35696985391564e-05,
+      "loss": 0.8235,
+      "step": 216220
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35689368897284e-05,
+      "loss": 0.863,
+      "step": 216230
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.35681752403004e-05,
+      "loss": 0.8809,
+      "step": 216240
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.356741359087239e-05,
+      "loss": 0.8812,
+      "step": 216250
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.3566651941444396e-05,
+      "loss": 0.9149,
+      "step": 216260
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356589029201639e-05,
+      "loss": 0.8619,
+      "step": 216270
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356512864258839e-05,
+      "loss": 0.8512,
+      "step": 216280
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356436699316039e-05,
+      "loss": 0.8662,
+      "step": 216290
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356360534373239e-05,
+      "loss": 0.897,
+      "step": 216300
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356284369430438e-05,
+      "loss": 0.7729,
+      "step": 216310
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356208204487639e-05,
+      "loss": 0.803,
+      "step": 216320
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3561320395448385e-05,
+      "loss": 0.8548,
+      "step": 216330
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.356055874602038e-05,
+      "loss": 0.8697,
+      "step": 216340
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.355979709659238e-05,
+      "loss": 0.9331,
+      "step": 216350
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3559035447164384e-05,
+      "loss": 0.9075,
+      "step": 216360
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3558273797736375e-05,
+      "loss": 0.8107,
+      "step": 216370
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.355751214830838e-05,
+      "loss": 0.8181,
+      "step": 216380
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3556750498880377e-05,
+      "loss": 0.8757,
+      "step": 216390
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3555988849452374e-05,
+      "loss": 0.7841,
+      "step": 216400
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.355522720002438e-05,
+      "loss": 0.9007,
+      "step": 216410
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3554465550596376e-05,
+      "loss": 0.8297,
+      "step": 216420
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.355370390116837e-05,
+      "loss": 0.9666,
+      "step": 216430
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.355294225174037e-05,
+      "loss": 0.8765,
+      "step": 216440
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3552180602312375e-05,
+      "loss": 0.7628,
+      "step": 216450
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.3551418952884366e-05,
+      "loss": 0.9995,
+      "step": 216460
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.355065730345637e-05,
+      "loss": 0.8158,
+      "step": 216470
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354989565402837e-05,
+      "loss": 0.8456,
+      "step": 216480
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3549134004600365e-05,
+      "loss": 0.815,
+      "step": 216490
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354837235517236e-05,
+      "loss": 0.8559,
+      "step": 216500
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354761070574436e-05,
+      "loss": 0.8707,
+      "step": 216510
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354684905631636e-05,
+      "loss": 0.8941,
+      "step": 216520
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354608740688836e-05,
+      "loss": 0.825,
+      "step": 216530
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354532575746035e-05,
+      "loss": 0.7629,
+      "step": 216540
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3544564108032356e-05,
+      "loss": 0.8756,
+      "step": 216550
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3543802458604354e-05,
+      "loss": 0.8489,
+      "step": 216560
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354304080917635e-05,
+      "loss": 0.8481,
+      "step": 216570
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3542279159748355e-05,
+      "loss": 0.8933,
+      "step": 216580
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354151751032035e-05,
+      "loss": 0.8223,
+      "step": 216590
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.354075586089235e-05,
+      "loss": 0.9154,
+      "step": 216600
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.353999421146435e-05,
+      "loss": 0.7999,
+      "step": 216610
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.353923256203635e-05,
+      "loss": 0.7744,
+      "step": 216620
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.353847091260834e-05,
+      "loss": 0.8613,
+      "step": 216630
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.353770926318035e-05,
+      "loss": 0.899,
+      "step": 216640
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3536947613752345e-05,
+      "loss": 0.9305,
+      "step": 216650
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.353618596432434e-05,
+      "loss": 0.8166,
+      "step": 216660
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.353542431489634e-05,
+      "loss": 0.9475,
+      "step": 216670
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3534662665468344e-05,
+      "loss": 0.8968,
+      "step": 216680
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3533901016040334e-05,
+      "loss": 0.7584,
+      "step": 216690
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.353313936661234e-05,
+      "loss": 0.9324,
+      "step": 216700
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3532377717184336e-05,
+      "loss": 0.8886,
+      "step": 216710
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3531616067756334e-05,
+      "loss": 0.797,
+      "step": 216720
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.353085441832833e-05,
+      "loss": 0.8366,
+      "step": 216730
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3530092768900335e-05,
+      "loss": 0.9519,
+      "step": 216740
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3529331119472326e-05,
+      "loss": 0.9666,
+      "step": 216750
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352856947004433e-05,
+      "loss": 0.9598,
+      "step": 216760
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352780782061633e-05,
+      "loss": 0.9004,
+      "step": 216770
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3527046171188325e-05,
+      "loss": 0.8831,
+      "step": 216780
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352628452176033e-05,
+      "loss": 0.8597,
+      "step": 216790
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352552287233233e-05,
+      "loss": 0.8357,
+      "step": 216800
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3524761222904324e-05,
+      "loss": 0.8873,
+      "step": 216810
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352399957347632e-05,
+      "loss": 0.8997,
+      "step": 216820
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3523237924048326e-05,
+      "loss": 0.8731,
+      "step": 216830
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352247627462032e-05,
+      "loss": 0.8737,
+      "step": 216840
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352171462519232e-05,
+      "loss": 0.8601,
+      "step": 216850
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.352095297576432e-05,
+      "loss": 0.9626,
+      "step": 216860
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.3520191326336316e-05,
+      "loss": 0.8277,
+      "step": 216870
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.351942967690831e-05,
+      "loss": 0.9195,
+      "step": 216880
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.351866802748032e-05,
+      "loss": 0.9305,
+      "step": 216890
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.351790637805231e-05,
+      "loss": 0.8876,
+      "step": 216900
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.351714472862431e-05,
+      "loss": 0.8279,
+      "step": 216910
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.351638307919631e-05,
+      "loss": 0.8343,
+      "step": 216920
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.351562142976831e-05,
+      "loss": 0.7826,
+      "step": 216930
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3514859780340305e-05,
+      "loss": 0.8138,
+      "step": 216940
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.35140981309123e-05,
+      "loss": 0.7966,
+      "step": 216950
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.35133364814843e-05,
+      "loss": 0.9356,
+      "step": 216960
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3512574832056304e-05,
+      "loss": 0.8573,
+      "step": 216970
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.35118131826283e-05,
+      "loss": 0.9013,
+      "step": 216980
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.35110515332003e-05,
+      "loss": 0.8984,
+      "step": 216990
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.35102898837723e-05,
+      "loss": 0.8549,
+      "step": 217000
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3509528234344294e-05,
+      "loss": 0.9083,
+      "step": 217010
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.35087665849163e-05,
+      "loss": 0.8588,
+      "step": 217020
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3508004935488296e-05,
+      "loss": 0.9047,
+      "step": 217030
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.350724328606029e-05,
+      "loss": 0.8689,
+      "step": 217040
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.350648163663229e-05,
+      "loss": 0.9411,
+      "step": 217050
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3505719987204295e-05,
+      "loss": 0.8025,
+      "step": 217060
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3504958337776286e-05,
+      "loss": 0.8398,
+      "step": 217070
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.350419668834829e-05,
+      "loss": 0.878,
+      "step": 217080
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.350343503892029e-05,
+      "loss": 0.8433,
+      "step": 217090
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3502673389492285e-05,
+      "loss": 0.7667,
+      "step": 217100
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.350191174006428e-05,
+      "loss": 0.8609,
+      "step": 217110
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.3501150090636286e-05,
+      "loss": 0.823,
+      "step": 217120
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.350038844120828e-05,
+      "loss": 0.8374,
+      "step": 217130
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349962679178028e-05,
+      "loss": 0.8634,
+      "step": 217140
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349886514235228e-05,
+      "loss": 0.9203,
+      "step": 217150
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3498103492924276e-05,
+      "loss": 0.8092,
+      "step": 217160
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3497341843496274e-05,
+      "loss": 0.9646,
+      "step": 217170
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349658019406828e-05,
+      "loss": 0.9225,
+      "step": 217180
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3495818544640275e-05,
+      "loss": 0.8652,
+      "step": 217190
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349505689521227e-05,
+      "loss": 0.9344,
+      "step": 217200
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349429524578428e-05,
+      "loss": 0.8918,
+      "step": 217210
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349353359635627e-05,
+      "loss": 0.8819,
+      "step": 217220
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349277194692827e-05,
+      "loss": 0.8126,
+      "step": 217230
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349201029750027e-05,
+      "loss": 0.9541,
+      "step": 217240
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.349124864807227e-05,
+      "loss": 0.8451,
+      "step": 217250
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3490486998644265e-05,
+      "loss": 0.8501,
+      "step": 217260
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.348972534921627e-05,
+      "loss": 1.0307,
+      "step": 217270
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.348896369978826e-05,
+      "loss": 0.7912,
+      "step": 217280
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3488202050360264e-05,
+      "loss": 0.8403,
+      "step": 217290
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.348744040093226e-05,
+      "loss": 0.9304,
+      "step": 217300
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.348667875150426e-05,
+      "loss": 0.793,
+      "step": 217310
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3485917102076256e-05,
+      "loss": 0.881,
+      "step": 217320
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.348515545264826e-05,
+      "loss": 0.8362,
+      "step": 217330
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.348439380322025e-05,
+      "loss": 0.8312,
+      "step": 217340
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3483632153792255e-05,
+      "loss": 0.8933,
+      "step": 217350
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.348287050436425e-05,
+      "loss": 0.9448,
+      "step": 217360
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.348210885493625e-05,
+      "loss": 0.8814,
+      "step": 217370
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3481347205508254e-05,
+      "loss": 0.8267,
+      "step": 217380
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.348058555608025e-05,
+      "loss": 0.8529,
+      "step": 217390
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347982390665225e-05,
+      "loss": 0.8678,
+      "step": 217400
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347906225722425e-05,
+      "loss": 0.9089,
+      "step": 217410
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3478300607796244e-05,
+      "loss": 0.881,
+      "step": 217420
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347753895836824e-05,
+      "loss": 0.8584,
+      "step": 217430
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3476777308940246e-05,
+      "loss": 0.8844,
+      "step": 217440
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347601565951224e-05,
+      "loss": 0.8389,
+      "step": 217450
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347525401008424e-05,
+      "loss": 0.8838,
+      "step": 217460
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347449236065624e-05,
+      "loss": 0.8226,
+      "step": 217470
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3473730711228236e-05,
+      "loss": 0.8281,
+      "step": 217480
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347296906180023e-05,
+      "loss": 0.8289,
+      "step": 217490
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347220741237224e-05,
+      "loss": 0.8963,
+      "step": 217500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347144576294423e-05,
+      "loss": 0.8204,
+      "step": 217510
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.347068411351623e-05,
+      "loss": 0.9059,
+      "step": 217520
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.346992246408823e-05,
+      "loss": 0.9098,
+      "step": 217530
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.346916081466023e-05,
+      "loss": 0.9246,
+      "step": 217540
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.3468399165232225e-05,
+      "loss": 0.9596,
+      "step": 217550
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.346763751580423e-05,
+      "loss": 0.9195,
+      "step": 217560
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.346687586637623e-05,
+      "loss": 0.8165,
+      "step": 217570
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.3466114216948224e-05,
+      "loss": 0.9557,
+      "step": 217580
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346535256752023e-05,
+      "loss": 0.8918,
+      "step": 217590
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346459091809222e-05,
+      "loss": 0.9362,
+      "step": 217600
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346382926866422e-05,
+      "loss": 0.8037,
+      "step": 217610
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346306761923622e-05,
+      "loss": 0.9121,
+      "step": 217620
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346230596980822e-05,
+      "loss": 0.8318,
+      "step": 217630
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.3461544320380216e-05,
+      "loss": 0.9187,
+      "step": 217640
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346078267095222e-05,
+      "loss": 0.8618,
+      "step": 217650
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.346002102152421e-05,
+      "loss": 0.8899,
+      "step": 217660
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.3459259372096215e-05,
+      "loss": 0.8877,
+      "step": 217670
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.345849772266821e-05,
+      "loss": 0.8689,
+      "step": 217680
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.345773607324021e-05,
+      "loss": 0.8196,
+      "step": 217690
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.345697442381221e-05,
+      "loss": 0.8522,
+      "step": 217700
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.345621277438421e-05,
+      "loss": 0.8687,
+      "step": 217710
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.34554511249562e-05,
+      "loss": 0.8181,
+      "step": 217720
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.3454689475528206e-05,
+      "loss": 0.9366,
+      "step": 217730
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.3453927826100204e-05,
+      "loss": 0.8869,
+      "step": 217740
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.34531661766722e-05,
+      "loss": 0.8832,
+      "step": 217750
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.34524045272442e-05,
+      "loss": 0.8972,
+      "step": 217760
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.34516428778162e-05,
+      "loss": 0.8363,
+      "step": 217770
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.34508812283882e-05,
+      "loss": 0.8866,
+      "step": 217780
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.34501195789602e-05,
+      "loss": 0.8825,
+      "step": 217790
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.34493579295322e-05,
+      "loss": 0.8035,
+      "step": 217800
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344859628010419e-05,
+      "loss": 0.8413,
+      "step": 217810
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.34478346306762e-05,
+      "loss": 0.8434,
+      "step": 217820
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3447072981248195e-05,
+      "loss": 0.9795,
+      "step": 217830
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344631133182019e-05,
+      "loss": 0.8309,
+      "step": 217840
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344554968239219e-05,
+      "loss": 0.7739,
+      "step": 217850
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3444788032964194e-05,
+      "loss": 0.8764,
+      "step": 217860
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3444026383536185e-05,
+      "loss": 0.7522,
+      "step": 217870
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344326473410819e-05,
+      "loss": 0.9847,
+      "step": 217880
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344250308468018e-05,
+      "loss": 0.9111,
+      "step": 217890
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3441741435252184e-05,
+      "loss": 0.8441,
+      "step": 217900
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344097978582418e-05,
+      "loss": 0.9564,
+      "step": 217910
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.344021813639618e-05,
+      "loss": 1.0132,
+      "step": 217920
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3439456486968176e-05,
+      "loss": 0.9186,
+      "step": 217930
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.343869483754018e-05,
+      "loss": 0.9545,
+      "step": 217940
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.343793318811218e-05,
+      "loss": 0.9565,
+      "step": 217950
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3437171538684175e-05,
+      "loss": 0.9376,
+      "step": 217960
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.343640988925617e-05,
+      "loss": 0.8893,
+      "step": 217970
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.343564823982817e-05,
+      "loss": 0.8341,
+      "step": 217980
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.3434886590400174e-05,
+      "loss": 0.8672,
+      "step": 217990
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.343412494097217e-05,
+      "loss": 0.9655,
+      "step": 218000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.343336329154417e-05,
+      "loss": 0.8301,
+      "step": 218010
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.343260164211617e-05,
+      "loss": 0.7915,
+      "step": 218020
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.343183999268817e-05,
+      "loss": 0.8086,
+      "step": 218030
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.343107834326016e-05,
+      "loss": 0.8696,
+      "step": 218040
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3430316693832166e-05,
+      "loss": 0.8727,
+      "step": 218050
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3429555044404164e-05,
+      "loss": 0.988,
+      "step": 218060
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342879339497616e-05,
+      "loss": 0.9401,
+      "step": 218070
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342803174554816e-05,
+      "loss": 0.8914,
+      "step": 218080
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342727009612016e-05,
+      "loss": 0.8841,
+      "step": 218090
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342650844669215e-05,
+      "loss": 0.7955,
+      "step": 218100
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342574679726416e-05,
+      "loss": 0.9103,
+      "step": 218110
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3424985147836155e-05,
+      "loss": 0.817,
+      "step": 218120
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342422349840815e-05,
+      "loss": 0.9253,
+      "step": 218130
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342346184898015e-05,
+      "loss": 0.8858,
+      "step": 218140
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3422700199552154e-05,
+      "loss": 0.8889,
+      "step": 218150
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342193855012415e-05,
+      "loss": 0.8318,
+      "step": 218160
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.342117690069615e-05,
+      "loss": 0.8653,
+      "step": 218170
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3420415251268153e-05,
+      "loss": 0.9213,
+      "step": 218180
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3419653601840144e-05,
+      "loss": 0.8419,
+      "step": 218190
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.341889195241215e-05,
+      "loss": 0.8308,
+      "step": 218200
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.3418130302984146e-05,
+      "loss": 0.871,
+      "step": 218210
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.341736865355614e-05,
+      "loss": 0.9191,
+      "step": 218220
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.341660700412814e-05,
+      "loss": 0.8639,
+      "step": 218230
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3415845354700145e-05,
+      "loss": 0.8081,
+      "step": 218240
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3415083705272136e-05,
+      "loss": 0.7993,
+      "step": 218250
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.341432205584414e-05,
+      "loss": 0.9269,
+      "step": 218260
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.341356040641614e-05,
+      "loss": 0.8353,
+      "step": 218270
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3412798756988135e-05,
+      "loss": 0.8964,
+      "step": 218280
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.341203710756013e-05,
+      "loss": 0.948,
+      "step": 218290
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3411275458132137e-05,
+      "loss": 0.8553,
+      "step": 218300
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.341051380870413e-05,
+      "loss": 0.8031,
+      "step": 218310
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340975215927613e-05,
+      "loss": 1.0281,
+      "step": 218320
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340899050984813e-05,
+      "loss": 0.8678,
+      "step": 218330
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3408228860420126e-05,
+      "loss": 0.8827,
+      "step": 218340
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3407467210992124e-05,
+      "loss": 0.9076,
+      "step": 218350
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340670556156412e-05,
+      "loss": 0.9577,
+      "step": 218360
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3405943912136126e-05,
+      "loss": 0.9306,
+      "step": 218370
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340518226270812e-05,
+      "loss": 0.8773,
+      "step": 218380
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340442061328012e-05,
+      "loss": 0.8599,
+      "step": 218390
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340365896385212e-05,
+      "loss": 0.8191,
+      "step": 218400
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340289731442412e-05,
+      "loss": 0.8064,
+      "step": 218410
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340213566499611e-05,
+      "loss": 0.9053,
+      "step": 218420
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.340137401556812e-05,
+      "loss": 0.9854,
+      "step": 218430
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.3400612366140115e-05,
+      "loss": 0.8782,
+      "step": 218440
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.339985071671211e-05,
+      "loss": 0.9488,
+      "step": 218450
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.339908906728411e-05,
+      "loss": 0.8525,
+      "step": 218460
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3398327417856114e-05,
+      "loss": 0.8918,
+      "step": 218470
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3397565768428105e-05,
+      "loss": 0.8597,
+      "step": 218480
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.339680411900011e-05,
+      "loss": 1.04,
+      "step": 218490
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3396042469572106e-05,
+      "loss": 0.7427,
+      "step": 218500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3395280820144104e-05,
+      "loss": 0.9602,
+      "step": 218510
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.33945191707161e-05,
+      "loss": 0.8645,
+      "step": 218520
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3393757521288105e-05,
+      "loss": 1.0797,
+      "step": 218530
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.33929958718601e-05,
+      "loss": 1.004,
+      "step": 218540
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.33922342224321e-05,
+      "loss": 0.824,
+      "step": 218550
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.33914725730041e-05,
+      "loss": 0.9154,
+      "step": 218560
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3390710923576095e-05,
+      "loss": 0.877,
+      "step": 218570
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.33899492741481e-05,
+      "loss": 0.9031,
+      "step": 218580
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.33891876247201e-05,
+      "loss": 0.8414,
+      "step": 218590
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3388425975292094e-05,
+      "loss": 0.9586,
+      "step": 218600
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.338766432586409e-05,
+      "loss": 0.8747,
+      "step": 218610
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3386902676436096e-05,
+      "loss": 0.8683,
+      "step": 218620
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.338614102700809e-05,
+      "loss": 0.8816,
+      "step": 218630
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.338537937758009e-05,
+      "loss": 0.8796,
+      "step": 218640
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.338461772815209e-05,
+      "loss": 0.8636,
+      "step": 218650
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.3383856078724086e-05,
+      "loss": 0.8825,
+      "step": 218660
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3383094429296084e-05,
+      "loss": 0.8548,
+      "step": 218670
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.338233277986809e-05,
+      "loss": 0.8788,
+      "step": 218680
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.338157113044008e-05,
+      "loss": 0.8835,
+      "step": 218690
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.338080948101208e-05,
+      "loss": 0.8564,
+      "step": 218700
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.338004783158408e-05,
+      "loss": 0.8261,
+      "step": 218710
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337928618215608e-05,
+      "loss": 0.7926,
+      "step": 218720
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3378524532728075e-05,
+      "loss": 0.783,
+      "step": 218730
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337776288330008e-05,
+      "loss": 0.8192,
+      "step": 218740
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337700123387208e-05,
+      "loss": 0.9021,
+      "step": 218750
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3376239584444074e-05,
+      "loss": 0.858,
+      "step": 218760
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337547793501607e-05,
+      "loss": 0.8277,
+      "step": 218770
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337471628558807e-05,
+      "loss": 0.8834,
+      "step": 218780
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3373954636160073e-05,
+      "loss": 0.8695,
+      "step": 218790
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3373192986732064e-05,
+      "loss": 0.7745,
+      "step": 218800
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337243133730407e-05,
+      "loss": 0.7946,
+      "step": 218810
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3371669687876066e-05,
+      "loss": 0.8616,
+      "step": 218820
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337090803844806e-05,
+      "loss": 1.0445,
+      "step": 218830
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.337014638902006e-05,
+      "loss": 0.8394,
+      "step": 218840
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3369384739592065e-05,
+      "loss": 0.762,
+      "step": 218850
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.3368623090164056e-05,
+      "loss": 0.8821,
+      "step": 218860
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.336786144073606e-05,
+      "loss": 0.8496,
+      "step": 218870
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.336709979130806e-05,
+      "loss": 0.9084,
+      "step": 218880
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3366338141880055e-05,
+      "loss": 0.8813,
+      "step": 218890
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.336557649245205e-05,
+      "loss": 0.883,
+      "step": 218900
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.336481484302406e-05,
+      "loss": 0.8145,
+      "step": 218910
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3364053193596054e-05,
+      "loss": 0.8616,
+      "step": 218920
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.336329154416805e-05,
+      "loss": 0.8024,
+      "step": 218930
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.336252989474005e-05,
+      "loss": 0.8242,
+      "step": 218940
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3361768245312046e-05,
+      "loss": 0.9054,
+      "step": 218950
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.336100659588405e-05,
+      "loss": 0.899,
+      "step": 218960
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.336024494645605e-05,
+      "loss": 0.84,
+      "step": 218970
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3359483297028046e-05,
+      "loss": 0.9241,
+      "step": 218980
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335872164760004e-05,
+      "loss": 0.8678,
+      "step": 218990
+    },
+    {
+      "epoch": 10.0,
+      "eval_cer": 0.6213628984143057,
+      "eval_em": 0.007568590350047304,
+      "eval_f1": 0.007568590350047304,
+      "eval_loss": 0.8919129967689514,
+      "eval_runtime": 1011.3314,
+      "eval_samples_per_second": 10.452,
+      "eval_steps_per_second": 1.307,
+      "eval_wer": 0.9924314096499527,
+      "step": 218997
+    }
+  ],
+  "max_steps": 656970,
+  "num_train_epochs": 30,
+  "total_flos": 1.7110728299049062e+18,
+  "trial_name": null,
+  "trial_params": null
+}