diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17948 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "global_step": 29664,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.6626,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 10.6253,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 10.6097,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 10.5975,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 10.5705,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 10.4576,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 9.9928,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 9.0461,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 8.2289,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 7.7897,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011,
+      "loss": 7.3567,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012,
+      "loss": 7.0239,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 7.0109,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 6.9051,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015,
+      "loss": 6.9468,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016,
+      "loss": 7.0454,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017,
+      "loss": 6.7261,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 6.9753,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019,
+      "loss": 6.9379,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 6.7928,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00021,
+      "loss": 7.0586,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00022,
+      "loss": 6.7062,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00023,
+      "loss": 6.7424,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00024,
+      "loss": 6.5632,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00025,
+      "loss": 6.4557,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 6.6338,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00027,
+      "loss": 6.5024,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 6.4847,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029,
+      "loss": 6.4058,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003,
+      "loss": 6.1462,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00031,
+      "loss": 6.1794,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00032,
+      "loss": 6.165,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00033,
+      "loss": 6.0411,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00034,
+      "loss": 5.8371,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00035,
+      "loss": 5.7552,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 5.7949,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00037,
+      "loss": 5.6373,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00038,
+      "loss": 5.5568,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 5.6464,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004,
+      "loss": 5.6136,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00041,
+      "loss": 5.3152,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00042,
+      "loss": 5.2037,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00043,
+      "loss": 5.3338,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00044,
+      "loss": 5.2914,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 5.1617,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00046,
+      "loss": 5.0742,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047,
+      "loss": 5.0335,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048,
+      "loss": 5.0184,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00049,
+      "loss": 4.951,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 4.811,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004998633132859486,
+      "loss": 4.9258,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004997266265718973,
+      "loss": 4.9171,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004995899398578459,
+      "loss": 4.8119,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004994532531437944,
+      "loss": 4.6364,
+      "step": 540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000499316566429743,
+      "loss": 4.8717,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004991798797156917,
+      "loss": 4.7453,
+      "step": 560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004990431930016403,
+      "loss": 4.7988,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004989065062875889,
+      "loss": 4.5973,
+      "step": 580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004987698195735374,
+      "loss": 4.6734,
+      "step": 590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004986331328594861,
+      "loss": 4.4877,
+      "step": 600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004984964461454347,
+      "loss": 4.6588,
+      "step": 610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004983597594313833,
+      "loss": 4.3761,
+      "step": 620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004982230727173318,
+      "loss": 4.3939,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004980863860032805,
+      "loss": 4.4562,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004979496992892291,
+      "loss": 4.4264,
+      "step": 650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004978130125751777,
+      "loss": 4.3819,
+      "step": 660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004976763258611263,
+      "loss": 4.4243,
+      "step": 670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004975396391470749,
+      "loss": 4.33,
+      "step": 680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004974029524330235,
+      "loss": 4.1939,
+      "step": 690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004972662657189721,
+      "loss": 4.0569,
+      "step": 700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004971295790049207,
+      "loss": 4.1522,
+      "step": 710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004969928922908694,
+      "loss": 4.0689,
+      "step": 720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004968562055768179,
+      "loss": 3.9732,
+      "step": 730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004967195188627665,
+      "loss": 4.0417,
+      "step": 740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004965828321487151,
+      "loss": 3.9111,
+      "step": 750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004964461454346638,
+      "loss": 3.8938,
+      "step": 760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004963094587206124,
+      "loss": 3.9351,
+      "step": 770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000496172772006561,
+      "loss": 4.0446,
+      "step": 780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004960360852925096,
+      "loss": 3.9448,
+      "step": 790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004958993985784582,
+      "loss": 3.765,
+      "step": 800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004957627118644068,
+      "loss": 4.004,
+      "step": 810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004956260251503554,
+      "loss": 3.7195,
+      "step": 820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000495489338436304,
+      "loss": 3.81,
+      "step": 830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004953526517222527,
+      "loss": 3.538,
+      "step": 840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004952159650082012,
+      "loss": 3.6894,
+      "step": 850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004950792782941498,
+      "loss": 3.5949,
+      "step": 860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004949425915800984,
+      "loss": 3.7688,
+      "step": 870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004948059048660471,
+      "loss": 3.5947,
+      "step": 880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004946692181519956,
+      "loss": 3.4927,
+      "step": 890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004945325314379442,
+      "loss": 3.5899,
+      "step": 900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004943958447238928,
+      "loss": 3.3451,
+      "step": 910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004942591580098415,
+      "loss": 3.3756,
+      "step": 920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004941224712957901,
+      "loss": 3.5748,
+      "step": 930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004939857845817386,
+      "loss": 3.5213,
+      "step": 940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004938490978676872,
+      "loss": 3.4859,
+      "step": 950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004937124111536359,
+      "loss": 3.3436,
+      "step": 960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004935757244395845,
+      "loss": 3.3632,
+      "step": 970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004934390377255331,
+      "loss": 3.4325,
+      "step": 980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004933023510114816,
+      "loss": 3.215,
+      "step": 990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004931656642974303,
+      "loss": 3.3593,
+      "step": 1000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004930289775833789,
+      "loss": 3.1892,
+      "step": 1010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004928922908693275,
+      "loss": 3.1589,
+      "step": 1020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000492755604155276,
+      "loss": 3.238,
+      "step": 1030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004926189174412248,
+      "loss": 3.2319,
+      "step": 1040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004924822307271734,
+      "loss": 3.1466,
+      "step": 1050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004923455440131219,
+      "loss": 3.2193,
+      "step": 1060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004922088572990705,
+      "loss": 3.1091,
+      "step": 1070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004920721705850192,
+      "loss": 3.3262,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004919354838709678,
+      "loss": 3.1234,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004917987971569163,
+      "loss": 3.1121,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000491662110442865,
+      "loss": 3.1061,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004915254237288136,
+      "loss": 2.875,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004913887370147622,
+      "loss": 2.9009,
+      "step": 1130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004912520503007108,
+      "loss": 2.8254,
+      "step": 1140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004911153635866593,
+      "loss": 2.8775,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000490978676872608,
+      "loss": 2.8992,
+      "step": 1160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004908419901585566,
+      "loss": 2.878,
+      "step": 1170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004907053034445052,
+      "loss": 2.7918,
+      "step": 1180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004905686167304539,
+      "loss": 2.7717,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004904319300164024,
+      "loss": 2.7412,
+      "step": 1200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000490295243302351,
+      "loss": 2.7835,
+      "step": 1210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004901585565882996,
+      "loss": 2.7368,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004900218698742483,
+      "loss": 2.9304,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004898851831601968,
+      "loss": 2.7349,
+      "step": 1240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004897484964461454,
+      "loss": 2.74,
+      "step": 1250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000489611809732094,
+      "loss": 2.6362,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004894751230180427,
+      "loss": 2.7151,
+      "step": 1270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004893384363039913,
+      "loss": 2.7185,
+      "step": 1280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004892017495899398,
+      "loss": 2.5048,
+      "step": 1290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004890650628758885,
+      "loss": 2.6805,
+      "step": 1300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004889283761618372,
+      "loss": 2.4164,
+      "step": 1310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004887916894477857,
+      "loss": 2.5574,
+      "step": 1320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004886550027337343,
+      "loss": 2.6981,
+      "step": 1330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004885183160196829,
+      "loss": 2.6306,
+      "step": 1340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004883816293056316,
+      "loss": 2.5428,
+      "step": 1350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004882449425915801,
+      "loss": 2.6213,
+      "step": 1360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004881082558775287,
+      "loss": 2.4055,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00048797156916347733,
+      "loss": 2.3518,
+      "step": 1380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004878348824494259,
+      "loss": 2.386,
+      "step": 1390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048769819573537453,
+      "loss": 2.3242,
+      "step": 1400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004875615090213231,
+      "loss": 2.5691,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048742482230727173,
+      "loss": 2.3799,
+      "step": 1420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004872881355932203,
+      "loss": 2.4306,
+      "step": 1430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000487151448879169,
+      "loss": 2.3532,
+      "step": 1440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004870147621651176,
+      "loss": 2.524,
+      "step": 1450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004868780754510662,
+      "loss": 2.385,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004867413887370148,
+      "loss": 2.3943,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004866047020229634,
+      "loss": 2.5039,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000486468015308912,
+      "loss": 2.2678,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004863313285948606,
+      "loss": 2.3408,
+      "step": 1500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004861946418808092,
+      "loss": 2.3589,
+      "step": 1510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048605795516675783,
+      "loss": 2.364,
+      "step": 1520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004859212684527064,
+      "loss": 2.1347,
+      "step": 1530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00048578458173865503,
+      "loss": 2.1827,
+      "step": 1540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004856478950246036,
+      "loss": 2.2381,
+      "step": 1550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00048551120831055223,
+      "loss": 2.2331,
+      "step": 1560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004853745215965008,
+      "loss": 2.257,
+      "step": 1570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00048523783488244943,
+      "loss": 2.3163,
+      "step": 1580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00048510114816839806,
+      "loss": 2.2193,
+      "step": 1590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00048496446145434663,
+      "loss": 2.2293,
+      "step": 1600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00048482777474029526,
+      "loss": 2.3064,
+      "step": 1610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048469108802624383,
+      "loss": 2.2361,
+      "step": 1620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048455440131219246,
+      "loss": 2.0809,
+      "step": 1630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048441771459814103,
+      "loss": 2.1405,
+      "step": 1640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048428102788408966,
+      "loss": 2.2043,
+      "step": 1650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004841443411700383,
+      "loss": 2.2389,
+      "step": 1660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00048400765445598686,
+      "loss": 2.1123,
+      "step": 1670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004838709677419355,
+      "loss": 2.1638,
+      "step": 1680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048373428102788406,
+      "loss": 1.978,
+      "step": 1690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048359759431383274,
+      "loss": 2.1414,
+      "step": 1700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004834609075997813,
+      "loss": 2.3218,
+      "step": 1710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048332422088572994,
+      "loss": 2.0116,
+      "step": 1720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00048318753417167856,
+      "loss": 2.0731,
+      "step": 1730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00048305084745762714,
+      "loss": 1.9902,
+      "step": 1740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00048291416074357576,
+      "loss": 2.0367,
+      "step": 1750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00048277747402952434,
+      "loss": 1.8839,
+      "step": 1760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00048264078731547296,
+      "loss": 1.9899,
+      "step": 1770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00048250410060142153,
+      "loss": 2.0824,
+      "step": 1780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00048236741388737016,
+      "loss": 2.1043,
+      "step": 1790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004822307271733188,
+      "loss": 2.0569,
+      "step": 1800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00048209404045926736,
+      "loss": 1.997,
+      "step": 1810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000481957353745216,
+      "loss": 1.8754,
+      "step": 1820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00048182066703116456,
+      "loss": 2.1285,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004816839803171132,
+      "loss": 1.851,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004815472936030618,
+      "loss": 2.0535,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004814106068890104,
+      "loss": 2.109,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000481273920174959,
+      "loss": 1.9947,
+      "step": 1870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004811372334609076,
+      "loss": 1.9508,
+      "step": 1880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004810005467468562,
+      "loss": 1.9339,
+      "step": 1890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004808638600328048,
+      "loss": 1.9757,
+      "step": 1900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004807271733187534,
+      "loss": 1.9308,
+      "step": 1910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048059048660470204,
+      "loss": 1.82,
+      "step": 1920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004804537998906506,
+      "loss": 1.9896,
+      "step": 1930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048031711317659924,
+      "loss": 1.7951,
+      "step": 1940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004801804264625478,
+      "loss": 1.8121,
+      "step": 1950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004800437397484965,
+      "loss": 1.8122,
+      "step": 1960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047990705303444507,
+      "loss": 1.8128,
+      "step": 1970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004797703663203937,
+      "loss": 1.9425,
+      "step": 1980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004796336796063423,
+      "loss": 2.0493,
+      "step": 1990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004794969928922909,
+      "loss": 1.8792,
+      "step": 2000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004793603061782395,
+      "loss": 1.7329,
+      "step": 2010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004792236194641881,
+      "loss": 1.8292,
+      "step": 2020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004790869327501367,
+      "loss": 1.9251,
+      "step": 2030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004789502460360853,
+      "loss": 1.8402,
+      "step": 2040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004788135593220339,
+      "loss": 1.954,
+      "step": 2050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00047867687260798254,
+      "loss": 1.94,
+      "step": 2060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004785401858939311,
+      "loss": 1.7552,
+      "step": 2070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00047840349917987974,
+      "loss": 1.9425,
+      "step": 2080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004782668124658283,
+      "loss": 1.7902,
+      "step": 2090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00047813012575177694,
+      "loss": 1.7799,
+      "step": 2100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004779934390377255,
+      "loss": 1.8514,
+      "step": 2110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00047785675232367414,
+      "loss": 1.7081,
+      "step": 2120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00047772006560962277,
+      "loss": 1.6707,
+      "step": 2130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00047758337889557134,
+      "loss": 1.7587,
+      "step": 2140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00047744669218151997,
+      "loss": 1.8205,
+      "step": 2150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00047731000546746854,
+      "loss": 1.721,
+      "step": 2160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00047717331875341717,
+      "loss": 1.6134,
+      "step": 2170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00047703663203936574,
+      "loss": 1.7683,
+      "step": 2180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00047689994532531437,
+      "loss": 1.7747,
+      "step": 2190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000476763258611263,
+      "loss": 1.6735,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00047662657189721157,
+      "loss": 1.8395,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0004764898851831602,
+      "loss": 1.8184,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0004763531984691088,
+      "loss": 1.8181,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00047621651175505745,
+      "loss": 1.8501,
+      "step": 2240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000476079825041006,
+      "loss": 1.663,
+      "step": 2250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00047594313832695465,
+      "loss": 1.7441,
+      "step": 2260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0004758064516129033,
+      "loss": 1.6841,
+      "step": 2270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00047566976489885185,
+      "loss": 1.8606,
+      "step": 2280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0004755330781848005,
+      "loss": 1.6687,
+      "step": 2290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00047539639147074905,
+      "loss": 1.7421,
+      "step": 2300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0004752597047566977,
+      "loss": 1.6962,
+      "step": 2310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00047512301804264625,
+      "loss": 1.7045,
+      "step": 2320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00047498633132859487,
+      "loss": 1.7425,
+      "step": 2330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0004748496446145435,
+      "loss": 1.6282,
+      "step": 2340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00047471295790049207,
+      "loss": 1.7722,
+      "step": 2350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0004745762711864407,
+      "loss": 1.7784,
+      "step": 2360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00047443958447238927,
+      "loss": 1.6109,
+      "step": 2370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0004743028977583379,
+      "loss": 1.5979,
+      "step": 2380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0004741662110442865,
+      "loss": 1.7631,
+      "step": 2390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004740295243302351,
+      "loss": 1.7131,
+      "step": 2400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004738928376161837,
+      "loss": 1.757,
+      "step": 2410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004737561509021323,
+      "loss": 1.5586,
+      "step": 2420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0004736194641880809,
+      "loss": 1.6299,
+      "step": 2430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0004734827774740295,
+      "loss": 1.6556,
+      "step": 2440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0004733460907599781,
+      "loss": 1.6007,
+      "step": 2450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00047320940404592675,
+      "loss": 1.6483,
+      "step": 2460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004730727173318753,
+      "loss": 1.634,
+      "step": 2470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00047293603061782395,
+      "loss": 1.5016,
+      "step": 2480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004727993439037726,
+      "loss": 1.765,
+      "step": 2490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004726626571897212,
+      "loss": 1.6249,
+      "step": 2500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004725259704756698,
+      "loss": 1.6578,
+      "step": 2510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004723892837616184,
+      "loss": 1.4478,
+      "step": 2520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047225259704756703,
+      "loss": 1.5294,
+      "step": 2530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004721159103335156,
+      "loss": 1.5331,
+      "step": 2540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047197922361946423,
+      "loss": 1.4759,
+      "step": 2550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004718425369054128,
+      "loss": 1.5038,
+      "step": 2560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047170585019136143,
+      "loss": 1.5278,
+      "step": 2570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00047156916347731,
+      "loss": 1.4657,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00047143247676325863,
+      "loss": 1.4285,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00047129579004920726,
+      "loss": 1.5785,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00047115910333515583,
+      "loss": 1.5626,
+      "step": 2610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00047102241662110445,
+      "loss": 1.4488,
+      "step": 2620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00047088572990705303,
+      "loss": 1.5122,
+      "step": 2630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00047074904319300165,
+      "loss": 1.4395,
+      "step": 2640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004706123564789502,
+      "loss": 1.4602,
+      "step": 2650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00047047566976489885,
+      "loss": 1.4403,
+      "step": 2660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004703389830508475,
+      "loss": 1.4329,
+      "step": 2670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00047020229633679605,
+      "loss": 1.523,
+      "step": 2680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004700656096227447,
+      "loss": 1.4406,
+      "step": 2690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046992892290869325,
+      "loss": 1.5166,
+      "step": 2700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004697922361946419,
+      "loss": 1.481,
+      "step": 2710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046965554948059045,
+      "loss": 1.4732,
+      "step": 2720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004695188627665391,
+      "loss": 1.6509,
+      "step": 2730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004693821760524877,
+      "loss": 1.5027,
+      "step": 2740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004692454893384363,
+      "loss": 1.454,
+      "step": 2750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00046910880262438496,
+      "loss": 1.4753,
+      "step": 2760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00046897211591033353,
+      "loss": 1.3976,
+      "step": 2770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00046883542919628216,
+      "loss": 1.4586,
+      "step": 2780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00046869874248223073,
+      "loss": 1.326,
+      "step": 2790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00046856205576817936,
+      "loss": 1.5128,
+      "step": 2800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000468425369054128,
+      "loss": 1.4267,
+      "step": 2810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00046828868234007656,
+      "loss": 1.5301,
+      "step": 2820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004681519956260252,
+      "loss": 1.4057,
+      "step": 2830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00046801530891197376,
+      "loss": 1.4616,
+      "step": 2840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004678786221979224,
+      "loss": 1.542,
+      "step": 2850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00046774193548387096,
+      "loss": 1.5282,
+      "step": 2860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004676052487698196,
+      "loss": 1.4366,
+      "step": 2870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0004674685620557682,
+      "loss": 1.3636,
+      "step": 2880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0004673318753417168,
+      "loss": 1.3394,
+      "step": 2890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0004671951886276654,
+      "loss": 1.4072,
+      "step": 2900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000467058501913614,
+      "loss": 1.6461,
+      "step": 2910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0004669218151995626,
+      "loss": 1.468,
+      "step": 2920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00046678512848551124,
+      "loss": 1.3194,
+      "step": 2930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0004666484417714598,
+      "loss": 1.3899,
+      "step": 2940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00046651175505740844,
+      "loss": 1.4353,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000466375068343357,
+      "loss": 1.3837,
+      "step": 2960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00046623838162930563,
+      "loss": 1.468,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0004661016949152542,
+      "loss": 1.2956,
+      "step": 2980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00046596500820120283,
+      "loss": 1.2953,
+      "step": 2990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00046582832148715146,
+      "loss": 1.3957,
+      "step": 3000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00046569163477310003,
+      "loss": 1.2778,
+      "step": 3010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0004655549480590487,
+      "loss": 1.3405,
+      "step": 3020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0004654182613449973,
+      "loss": 1.4767,
+      "step": 3030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0004652815746309459,
+      "loss": 1.3844,
+      "step": 3040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0004651448879168945,
+      "loss": 1.3703,
+      "step": 3050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0004650082012028431,
+      "loss": 1.5844,
+      "step": 3060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00046487151448879174,
+      "loss": 1.3405,
+      "step": 3070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0004647348277747403,
+      "loss": 1.4116,
+      "step": 3080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00046459814106068894,
+      "loss": 1.332,
+      "step": 3090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0004644614543466375,
+      "loss": 1.5187,
+      "step": 3100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00046432476763258614,
+      "loss": 1.4596,
+      "step": 3110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0004641880809185347,
+      "loss": 1.2941,
+      "step": 3120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00046405139420448334,
+      "loss": 1.5696,
+      "step": 3130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046391470749043197,
+      "loss": 1.325,
+      "step": 3140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046377802077638054,
+      "loss": 1.4449,
+      "step": 3150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046364133406232917,
+      "loss": 1.438,
+      "step": 3160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046350464734827774,
+      "loss": 1.3622,
+      "step": 3170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00046336796063422636,
+      "loss": 1.3304,
+      "step": 3180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00046323127392017494,
+      "loss": 1.5733,
+      "step": 3190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00046309458720612356,
+      "loss": 1.4687,
+      "step": 3200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0004629579004920722,
+      "loss": 1.397,
+      "step": 3210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00046282121377802076,
+      "loss": 1.4184,
+      "step": 3220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0004626845270639694,
+      "loss": 1.4533,
+      "step": 3230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00046254784034991796,
+      "loss": 1.4265,
+      "step": 3240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004624111536358666,
+      "loss": 1.2537,
+      "step": 3250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00046227446692181516,
+      "loss": 1.5875,
+      "step": 3260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004621377802077638,
+      "loss": 1.3558,
+      "step": 3270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004620010934937124,
+      "loss": 1.2115,
+      "step": 3280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00046186440677966104,
+      "loss": 1.4333,
+      "step": 3290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00046172772006560967,
+      "loss": 1.3109,
+      "step": 3300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00046159103335155824,
+      "loss": 1.413,
+      "step": 3310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00046145434663750687,
+      "loss": 1.3475,
+      "step": 3320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00046131765992345544,
+      "loss": 1.4351,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00046118097320940407,
+      "loss": 1.3709,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0004610442864953527,
+      "loss": 1.3424,
+      "step": 3350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00046090759978130127,
+      "loss": 1.4673,
+      "step": 3360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0004607709130672499,
+      "loss": 1.2657,
+      "step": 3370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00046063422635319847,
+      "loss": 1.5152,
+      "step": 3380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0004604975396391471,
+      "loss": 1.3713,
+      "step": 3390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00046036085292509567,
+      "loss": 1.3855,
+      "step": 3400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0004602241662110443,
+      "loss": 1.233,
+      "step": 3410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0004600874794969929,
+      "loss": 1.2909,
+      "step": 3420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004599507927829415,
+      "loss": 1.2725,
+      "step": 3430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004598141060688901,
+      "loss": 1.3344,
+      "step": 3440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004596774193548387,
+      "loss": 1.468,
+      "step": 3450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004595407326407873,
+      "loss": 1.3188,
+      "step": 3460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00045940404592673595,
+      "loss": 1.1457,
+      "step": 3470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0004592673592126845,
+      "loss": 1.2782,
+      "step": 3480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00045913067249863315,
+      "loss": 1.3477,
+      "step": 3490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0004589939857845817,
+      "loss": 1.4084,
+      "step": 3500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00045885729907053035,
+      "loss": 1.3303,
+      "step": 3510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0004587206123564789,
+      "loss": 1.2654,
+      "step": 3520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00045858392564242755,
+      "loss": 1.4258,
+      "step": 3530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00045844723892837617,
+      "loss": 1.4627,
+      "step": 3540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0004583105522143248,
+      "loss": 1.2783,
+      "step": 3550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0004581738655002734,
+      "loss": 1.3616,
+      "step": 3560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000458037178786222,
+      "loss": 1.3309,
+      "step": 3570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004579004920721706,
+      "loss": 1.1657,
+      "step": 3580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004577638053581192,
+      "loss": 1.1815,
+      "step": 3590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004576271186440678,
+      "loss": 1.4066,
+      "step": 3600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00045749043193001645,
+      "loss": 1.265,
+      "step": 3610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000457353745215965,
+      "loss": 1.3465,
+      "step": 3620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00045721705850191365,
+      "loss": 1.3139,
+      "step": 3630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0004570803717878622,
+      "loss": 1.3194,
+      "step": 3640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00045694368507381085,
+      "loss": 1.2924,
+      "step": 3650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0004568069983597594,
+      "loss": 1.3546,
+      "step": 3660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00045667031164570805,
+      "loss": 1.3168,
+      "step": 3670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0004565336249316567,
+      "loss": 1.0781,
+      "step": 3680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00045639693821760525,
+      "loss": 1.1993,
+      "step": 3690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004562602515035539,
+      "loss": 1.2988,
+      "step": 3700
+    },
+    {
+      "epoch": 1.0,
+      "eval_Bleu_1": 0.004715127701374937,
+      "eval_Bleu_2": 2.0649447537679265e-11,
+      "eval_Bleu_3": 3.7454969682283256e-14,
+      "eval_Bleu_4": 1.6959164979952297e-15,
+      "eval_ROUGE_L": 0.006043075272001848,
+      "eval_cer": 0.7164889732729662,
+      "eval_em": 0.33048433048433046,
+      "eval_f1": 0.4933815354050705,
+      "eval_loss": 1.1681952476501465,
+      "eval_runtime": 315.5092,
+      "eval_samples_per_second": 13.35,
+      "eval_steps_per_second": 2.672,
+      "eval_wer": 0.8473970944309928,
+      "step": 3708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00045612356478950245,
+      "loss": 1.2305,
+      "step": 3710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004559868780754511,
+      "loss": 0.9752,
+      "step": 3720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00045585019136139965,
+      "loss": 0.9645,
+      "step": 3730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004557135046473483,
+      "loss": 1.035,
+      "step": 3740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004555768179332969,
+      "loss": 1.0669,
+      "step": 3750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004554401312192455,
+      "loss": 1.007,
+      "step": 3760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004553034445051941,
+      "loss": 1.1085,
+      "step": 3770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004551667577911427,
+      "loss": 0.8787,
+      "step": 3780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004550300710770913,
+      "loss": 1.0839,
+      "step": 3790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004548933843630399,
+      "loss": 0.973,
+      "step": 3800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004547566976489885,
+      "loss": 0.9509,
+      "step": 3810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004546200109349372,
+      "loss": 0.9264,
+      "step": 3820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00045448332422088575,
+      "loss": 1.1141,
+      "step": 3830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004543466375068344,
+      "loss": 1.1239,
+      "step": 3840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00045420995079278295,
+      "loss": 1.077,
+      "step": 3850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004540732640787316,
+      "loss": 1.0122,
+      "step": 3860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00045393657736468015,
+      "loss": 1.0317,
+      "step": 3870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004537998906506288,
+      "loss": 1.0726,
+      "step": 3880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004536632039365774,
+      "loss": 0.9732,
+      "step": 3890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000453526517222526,
+      "loss": 1.0272,
+      "step": 3900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004533898305084746,
+      "loss": 1.2166,
+      "step": 3910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004532531437944232,
+      "loss": 0.9944,
+      "step": 3920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004531164570803718,
+      "loss": 1.1382,
+      "step": 3930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004529797703663204,
+      "loss": 1.0292,
+      "step": 3940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000452843083652269,
+      "loss": 0.9297,
+      "step": 3950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00045270639693821763,
+      "loss": 0.9691,
+      "step": 3960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004525697102241662,
+      "loss": 1.153,
+      "step": 3970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00045243302351011483,
+      "loss": 0.9683,
+      "step": 3980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004522963367960634,
+      "loss": 1.1609,
+      "step": 3990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045215965008201203,
+      "loss": 1.0564,
+      "step": 4000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045202296336796066,
+      "loss": 0.963,
+      "step": 4010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045188627665390923,
+      "loss": 0.9759,
+      "step": 4020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045174958993985786,
+      "loss": 0.9754,
+      "step": 4030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045161290322580643,
+      "loss": 1.1506,
+      "step": 4040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045147621651175506,
+      "loss": 0.9262,
+      "step": 4050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045133952979770363,
+      "loss": 0.9023,
+      "step": 4060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00045120284308365226,
+      "loss": 0.9211,
+      "step": 4070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00045106615636960094,
+      "loss": 0.981,
+      "step": 4080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004509294696555495,
+      "loss": 1.0214,
+      "step": 4090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00045079278294149814,
+      "loss": 0.9522,
+      "step": 4100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004506560962274467,
+      "loss": 1.0802,
+      "step": 4110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00045051940951339534,
+      "loss": 1.1325,
+      "step": 4120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004503827227993439,
+      "loss": 1.0782,
+      "step": 4130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00045024603608529254,
+      "loss": 0.9205,
+      "step": 4140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00045010934937124116,
+      "loss": 0.8969,
+      "step": 4150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00044997266265718973,
+      "loss": 0.9126,
+      "step": 4160
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00044983597594313836,
+      "loss": 0.9975,
+      "step": 4170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00044969928922908693,
+      "loss": 1.0562,
+      "step": 4180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00044956260251503556,
+      "loss": 1.1185,
+      "step": 4190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00044942591580098413,
+      "loss": 1.0614,
+      "step": 4200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00044928922908693276,
+      "loss": 0.9441,
+      "step": 4210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004491525423728814,
+      "loss": 1.0427,
+      "step": 4220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00044901585565882996,
+      "loss": 0.9571,
+      "step": 4230
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004488791689447786,
+      "loss": 1.0209,
+      "step": 4240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00044874248223072716,
+      "loss": 0.9572,
+      "step": 4250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004486057955166758,
+      "loss": 1.0203,
+      "step": 4260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00044846910880262436,
+      "loss": 0.8715,
+      "step": 4270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000448332422088573,
+      "loss": 0.8995,
+      "step": 4280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004481957353745216,
+      "loss": 0.9505,
+      "step": 4290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004480590486604702,
+      "loss": 0.9166,
+      "step": 4300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004479223619464188,
+      "loss": 1.0447,
+      "step": 4310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0004477856752323674,
+      "loss": 1.0449,
+      "step": 4320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000447648988518316,
+      "loss": 1.0228,
+      "step": 4330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0004475123018042646,
+      "loss": 1.0489,
+      "step": 4340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00044737561509021327,
+      "loss": 0.9268,
+      "step": 4350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0004472389283761619,
+      "loss": 0.9601,
+      "step": 4360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00044710224166211046,
+      "loss": 1.1113,
+      "step": 4370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0004469655549480591,
+      "loss": 1.1455,
+      "step": 4380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00044682886823400766,
+      "loss": 0.951,
+      "step": 4390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0004466921815199563,
+      "loss": 0.9629,
+      "step": 4400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00044655549480590486,
+      "loss": 0.9886,
+      "step": 4410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0004464188080918535,
+      "loss": 0.9926,
+      "step": 4420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0004462821213778021,
+      "loss": 0.9281,
+      "step": 4430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0004461454346637507,
+      "loss": 0.9411,
+      "step": 4440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0004460087479496993,
+      "loss": 1.061,
+      "step": 4450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0004458720612356479,
+      "loss": 0.9652,
+      "step": 4460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0004457353745215965,
+      "loss": 1.0107,
+      "step": 4470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0004455986878075451,
+      "loss": 1.0734,
+      "step": 4480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0004454620010934937,
+      "loss": 0.9813,
+      "step": 4490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00044532531437944234,
+      "loss": 0.998,
+      "step": 4500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004451886276653909,
+      "loss": 1.0986,
+      "step": 4510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00044505194095133954,
+      "loss": 0.9396,
+      "step": 4520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004449152542372881,
+      "loss": 0.9931,
+      "step": 4530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00044477856752323674,
+      "loss": 1.0652,
+      "step": 4540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00044464188080918537,
+      "loss": 1.0184,
+      "step": 4550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00044450519409513394,
+      "loss": 0.9795,
+      "step": 4560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00044436850738108257,
+      "loss": 1.0246,
+      "step": 4570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00044423182066703114,
+      "loss": 0.9178,
+      "step": 4580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00044409513395297977,
+      "loss": 1.0184,
+      "step": 4590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00044395844723892834,
+      "loss": 0.9148,
+      "step": 4600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000443821760524877,
+      "loss": 0.9315,
+      "step": 4610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00044368507381082565,
+      "loss": 0.9787,
+      "step": 4620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004435483870967742,
+      "loss": 1.036,
+      "step": 4630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00044341170038272285,
+      "loss": 1.0882,
+      "step": 4640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004432750136686714,
+      "loss": 1.0395,
+      "step": 4650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00044313832695462005,
+      "loss": 1.0066,
+      "step": 4660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0004430016402405686,
+      "loss": 1.0916,
+      "step": 4670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00044286495352651725,
+      "loss": 1.0187,
+      "step": 4680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00044272826681246587,
+      "loss": 1.0139,
+      "step": 4690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00044259158009841445,
+      "loss": 1.0061,
+      "step": 4700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00044245489338436307,
+      "loss": 1.0572,
+      "step": 4710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00044231820667031164,
+      "loss": 0.9492,
+      "step": 4720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00044218151995626027,
+      "loss": 0.8384,
+      "step": 4730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00044204483324220884,
+      "loss": 0.9791,
+      "step": 4740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00044190814652815747,
+      "loss": 0.9033,
+      "step": 4750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004417714598141061,
+      "loss": 0.8543,
+      "step": 4760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00044163477310005467,
+      "loss": 0.988,
+      "step": 4770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0004414980863860033,
+      "loss": 1.0482,
+      "step": 4780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00044136139967195187,
+      "loss": 0.9745,
+      "step": 4790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0004412247129579005,
+      "loss": 0.9758,
+      "step": 4800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00044108802624384907,
+      "loss": 1.0962,
+      "step": 4810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004409513395297977,
+      "loss": 1.0719,
+      "step": 4820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004408146528157463,
+      "loss": 0.8618,
+      "step": 4830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004406779661016949,
+      "loss": 0.8711,
+      "step": 4840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004405412793876435,
+      "loss": 0.9844,
+      "step": 4850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004404045926735921,
+      "loss": 1.0221,
+      "step": 4860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004402679059595407,
+      "loss": 1.072,
+      "step": 4870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00044013121924548935,
+      "loss": 1.0149,
+      "step": 4880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000439994532531438,
+      "loss": 1.1054,
+      "step": 4890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0004398578458173866,
+      "loss": 0.9934,
+      "step": 4900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0004397211591033352,
+      "loss": 0.9767,
+      "step": 4910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004395844723892838,
+      "loss": 0.9946,
+      "step": 4920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004394477856752324,
+      "loss": 1.0149,
+      "step": 4930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000439311098961181,
+      "loss": 1.1215,
+      "step": 4940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004391744122471296,
+      "loss": 1.0757,
+      "step": 4950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0004390377255330782,
+      "loss": 1.0171,
+      "step": 4960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00043890103881902683,
+      "loss": 0.9535,
+      "step": 4970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0004387643521049754,
+      "loss": 0.959,
+      "step": 4980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00043862766539092403,
+      "loss": 0.8601,
+      "step": 4990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0004384909786768726,
+      "loss": 0.8509,
+      "step": 5000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0004383542919628212,
+      "loss": 0.957,
+      "step": 5010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0004382176052487698,
+      "loss": 0.933,
+      "step": 5020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0004380809185347184,
+      "loss": 0.9639,
+      "step": 5030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00043794423182066705,
+      "loss": 0.9555,
+      "step": 5040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0004378075451066156,
+      "loss": 0.9016,
+      "step": 5050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00043767085839256425,
+      "loss": 0.9562,
+      "step": 5060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0004375341716785128,
+      "loss": 0.9805,
+      "step": 5070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00043739748496446145,
+      "loss": 0.9307,
+      "step": 5080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0004372607982504101,
+      "loss": 1.0861,
+      "step": 5090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00043712411153635865,
+      "loss": 0.9457,
+      "step": 5100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0004369874248223073,
+      "loss": 0.8683,
+      "step": 5110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00043685073810825585,
+      "loss": 0.9829,
+      "step": 5120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0004367140513942045,
+      "loss": 1.0282,
+      "step": 5130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00043657736468015305,
+      "loss": 0.9168,
+      "step": 5140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00043644067796610173,
+      "loss": 0.9717,
+      "step": 5150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00043630399125205036,
+      "loss": 0.9368,
+      "step": 5160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00043616730453799893,
+      "loss": 1.0143,
+      "step": 5170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00043603061782394756,
+      "loss": 0.9441,
+      "step": 5180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00043589393110989613,
+      "loss": 0.9788,
+      "step": 5190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00043575724439584476,
+      "loss": 0.9427,
+      "step": 5200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00043562055768179333,
+      "loss": 1.0207,
+      "step": 5210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00043548387096774196,
+      "loss": 0.9231,
+      "step": 5220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0004353471842536906,
+      "loss": 0.8983,
+      "step": 5230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00043521049753963916,
+      "loss": 1.0225,
+      "step": 5240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0004350738108255878,
+      "loss": 0.8657,
+      "step": 5250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00043493712411153636,
+      "loss": 0.8394,
+      "step": 5260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000434800437397485,
+      "loss": 0.9269,
+      "step": 5270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00043466375068343356,
+      "loss": 1.0183,
+      "step": 5280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004345270639693822,
+      "loss": 0.8344,
+      "step": 5290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004343903772553308,
+      "loss": 0.9371,
+      "step": 5300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004342536905412794,
+      "loss": 0.9828,
+      "step": 5310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000434117003827228,
+      "loss": 1.0401,
+      "step": 5320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004339803171131766,
+      "loss": 0.9944,
+      "step": 5330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004338436303991252,
+      "loss": 0.8523,
+      "step": 5340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004337069436850738,
+      "loss": 0.9157,
+      "step": 5350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0004335702569710224,
+      "loss": 0.9172,
+      "step": 5360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00043343357025697103,
+      "loss": 0.9604,
+      "step": 5370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0004332968835429196,
+      "loss": 0.9833,
+      "step": 5380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00043316019682886823,
+      "loss": 0.908,
+      "step": 5390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004330235101148168,
+      "loss": 0.9707,
+      "step": 5400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004328868234007655,
+      "loss": 1.0192,
+      "step": 5410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00043275013668671406,
+      "loss": 0.8414,
+      "step": 5420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004326134499726627,
+      "loss": 1.0844,
+      "step": 5430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0004324767632586113,
+      "loss": 0.8529,
+      "step": 5440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0004323400765445599,
+      "loss": 0.8889,
+      "step": 5450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0004322033898305085,
+      "loss": 1.0063,
+      "step": 5460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004320667031164571,
+      "loss": 0.9622,
+      "step": 5470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004319300164024057,
+      "loss": 0.9567,
+      "step": 5480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004317933296883543,
+      "loss": 0.8698,
+      "step": 5490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004316566429743029,
+      "loss": 0.9219,
+      "step": 5500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00043151995626025154,
+      "loss": 0.87,
+      "step": 5510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004313832695462001,
+      "loss": 0.8947,
+      "step": 5520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00043124658283214874,
+      "loss": 0.886,
+      "step": 5530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004311098961180973,
+      "loss": 0.972,
+      "step": 5540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00043097320940404594,
+      "loss": 0.9645,
+      "step": 5550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0004308365226899945,
+      "loss": 0.906,
+      "step": 5560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00043069983597594314,
+      "loss": 0.8833,
+      "step": 5570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00043056314926189176,
+      "loss": 0.8636,
+      "step": 5580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00043042646254784034,
+      "loss": 1.0244,
+      "step": 5590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00043028977583378896,
+      "loss": 0.9805,
+      "step": 5600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00043015308911973754,
+      "loss": 0.8972,
+      "step": 5610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00043001640240568616,
+      "loss": 0.9174,
+      "step": 5620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0004298797156916348,
+      "loss": 1.0851,
+      "step": 5630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00042974302897758336,
+      "loss": 0.9174,
+      "step": 5640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000429606342263532,
+      "loss": 0.9118,
+      "step": 5650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00042946965554948056,
+      "loss": 0.9208,
+      "step": 5660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00042933296883542924,
+      "loss": 0.8534,
+      "step": 5670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0004291962821213778,
+      "loss": 0.9325,
+      "step": 5680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00042905959540732644,
+      "loss": 0.9101,
+      "step": 5690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00042892290869327507,
+      "loss": 0.9682,
+      "step": 5700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00042878622197922364,
+      "loss": 0.9791,
+      "step": 5710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00042864953526517227,
+      "loss": 0.938,
+      "step": 5720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00042851284855112084,
+      "loss": 1.0013,
+      "step": 5730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00042837616183706947,
+      "loss": 0.9455,
+      "step": 5740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00042823947512301804,
+      "loss": 0.9269,
+      "step": 5750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00042810278840896667,
+      "loss": 0.9022,
+      "step": 5760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0004279661016949153,
+      "loss": 0.9125,
+      "step": 5770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00042782941498086387,
+      "loss": 0.8886,
+      "step": 5780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0004276927282668125,
+      "loss": 1.0339,
+      "step": 5790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00042755604155276107,
+      "loss": 0.881,
+      "step": 5800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004274193548387097,
+      "loss": 0.7516,
+      "step": 5810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00042728266812465827,
+      "loss": 0.8823,
+      "step": 5820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004271459814106069,
+      "loss": 0.9725,
+      "step": 5830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004270092946965555,
+      "loss": 0.978,
+      "step": 5840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0004268726079825041,
+      "loss": 1.0429,
+      "step": 5850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0004267359212684527,
+      "loss": 0.8862,
+      "step": 5860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0004265992345544013,
+      "loss": 0.9541,
+      "step": 5870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0004264625478403499,
+      "loss": 0.8402,
+      "step": 5880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0004263258611262985,
+      "loss": 0.8837,
+      "step": 5890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0004261891744122471,
+      "loss": 0.9571,
+      "step": 5900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00042605248769819574,
+      "loss": 0.9818,
+      "step": 5910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0004259158009841443,
+      "loss": 0.8562,
+      "step": 5920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00042577911427009294,
+      "loss": 0.9261,
+      "step": 5930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00042564242755604157,
+      "loss": 1.0236,
+      "step": 5940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0004255057408419902,
+      "loss": 0.932,
+      "step": 5950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00042536905412793877,
+      "loss": 0.8873,
+      "step": 5960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0004252323674138874,
+      "loss": 0.8528,
+      "step": 5970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000425095680699836,
+      "loss": 0.8803,
+      "step": 5980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004249589939857846,
+      "loss": 0.9814,
+      "step": 5990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004248223072717332,
+      "loss": 0.9202,
+      "step": 6000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004246856205576818,
+      "loss": 0.8905,
+      "step": 6010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004245489338436304,
+      "loss": 1.0263,
+      "step": 6020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000424412247129579,
+      "loss": 0.9521,
+      "step": 6030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0004242755604155276,
+      "loss": 0.9005,
+      "step": 6040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00042413887370147625,
+      "loss": 0.9133,
+      "step": 6050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0004240021869874248,
+      "loss": 0.8873,
+      "step": 6060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00042386550027337345,
+      "loss": 0.9688,
+      "step": 6070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000423728813559322,
+      "loss": 0.9342,
+      "step": 6080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00042359212684527065,
+      "loss": 0.871,
+      "step": 6090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004234554401312192,
+      "loss": 0.7895,
+      "step": 6100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00042331875341716785,
+      "loss": 0.8372,
+      "step": 6110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004231820667031165,
+      "loss": 0.9214,
+      "step": 6120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00042304537998906505,
+      "loss": 0.9022,
+      "step": 6130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004229086932750137,
+      "loss": 0.851,
+      "step": 6140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00042277200656096225,
+      "loss": 0.8247,
+      "step": 6150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004226353198469109,
+      "loss": 1.0589,
+      "step": 6160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004224986331328595,
+      "loss": 0.9376,
+      "step": 6170
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004223619464188081,
+      "loss": 0.8571,
+      "step": 6180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004222252597047567,
+      "loss": 0.9872,
+      "step": 6190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00042208857299070527,
+      "loss": 0.9051,
+      "step": 6200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00042195188627665395,
+      "loss": 0.895,
+      "step": 6210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004218151995626025,
+      "loss": 0.9938,
+      "step": 6220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00042167851284855115,
+      "loss": 0.9284,
+      "step": 6230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004215418261344998,
+      "loss": 0.8984,
+      "step": 6240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00042140513942044835,
+      "loss": 1.0252,
+      "step": 6250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000421268452706397,
+      "loss": 0.8839,
+      "step": 6260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00042113176599234555,
+      "loss": 0.7678,
+      "step": 6270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004209950792782942,
+      "loss": 0.8947,
+      "step": 6280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00042085839256424275,
+      "loss": 0.8449,
+      "step": 6290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004207217058501914,
+      "loss": 0.9152,
+      "step": 6300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00042058501913614,
+      "loss": 0.8691,
+      "step": 6310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004204483324220886,
+      "loss": 0.8582,
+      "step": 6320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004203116457080372,
+      "loss": 0.9852,
+      "step": 6330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004201749589939858,
+      "loss": 0.9016,
+      "step": 6340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004200382722799344,
+      "loss": 0.9155,
+      "step": 6350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000419901585565883,
+      "loss": 0.9502,
+      "step": 6360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004197648988518316,
+      "loss": 0.7541,
+      "step": 6370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00041962821213778023,
+      "loss": 0.8373,
+      "step": 6380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004194915254237288,
+      "loss": 0.9452,
+      "step": 6390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00041935483870967743,
+      "loss": 0.897,
+      "step": 6400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000419218151995626,
+      "loss": 0.8224,
+      "step": 6410
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00041908146528157463,
+      "loss": 0.8892,
+      "step": 6420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004189447785675232,
+      "loss": 0.8964,
+      "step": 6430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00041880809185347183,
+      "loss": 0.8986,
+      "step": 6440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00041867140513942046,
+      "loss": 0.7611,
+      "step": 6450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00041853471842536903,
+      "loss": 0.8113,
+      "step": 6460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004183980317113177,
+      "loss": 0.9327,
+      "step": 6470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004182613449972663,
+      "loss": 0.8745,
+      "step": 6480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004181246582832149,
+      "loss": 0.8833,
+      "step": 6490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004179879715691635,
+      "loss": 0.8765,
+      "step": 6500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004178512848551121,
+      "loss": 1.0384,
+      "step": 6510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00041771459814106074,
+      "loss": 0.8883,
+      "step": 6520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004175779114270093,
+      "loss": 0.8728,
+      "step": 6530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00041744122471295793,
+      "loss": 0.9931,
+      "step": 6540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004173045379989065,
+      "loss": 0.9192,
+      "step": 6550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00041716785128485513,
+      "loss": 0.9111,
+      "step": 6560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004170311645708037,
+      "loss": 0.9074,
+      "step": 6570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00041689447785675233,
+      "loss": 0.8986,
+      "step": 6580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00041675779114270096,
+      "loss": 0.8572,
+      "step": 6590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00041662110442864953,
+      "loss": 0.9254,
+      "step": 6600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00041648441771459816,
+      "loss": 0.796,
+      "step": 6610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00041634773100054673,
+      "loss": 0.7639,
+      "step": 6620
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00041621104428649536,
+      "loss": 0.9927,
+      "step": 6630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00041607435757244393,
+      "loss": 0.9595,
+      "step": 6640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00041593767085839256,
+      "loss": 0.8699,
+      "step": 6650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004158009841443412,
+      "loss": 0.9722,
+      "step": 6660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00041566429743028976,
+      "loss": 0.8815,
+      "step": 6670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004155276107162384,
+      "loss": 0.8739,
+      "step": 6680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00041539092400218696,
+      "loss": 0.9528,
+      "step": 6690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004152542372881356,
+      "loss": 0.7898,
+      "step": 6700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004151175505740842,
+      "loss": 0.9845,
+      "step": 6710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004149808638600328,
+      "loss": 0.9228,
+      "step": 6720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00041484417714598146,
+      "loss": 0.7807,
+      "step": 6730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00041470749043193004,
+      "loss": 1.0677,
+      "step": 6740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00041457080371787866,
+      "loss": 0.9074,
+      "step": 6750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00041443411700382724,
+      "loss": 0.9405,
+      "step": 6760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00041429743028977586,
+      "loss": 0.9472,
+      "step": 6770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004141607435757245,
+      "loss": 0.8346,
+      "step": 6780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00041402405686167306,
+      "loss": 0.916,
+      "step": 6790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004138873701476217,
+      "loss": 0.8767,
+      "step": 6800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00041375068343357026,
+      "loss": 0.8026,
+      "step": 6810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004136139967195189,
+      "loss": 0.9175,
+      "step": 6820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00041347731000546746,
+      "loss": 0.8865,
+      "step": 6830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004133406232914161,
+      "loss": 0.8086,
+      "step": 6840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004132039365773647,
+      "loss": 0.8606,
+      "step": 6850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004130672498633133,
+      "loss": 0.9845,
+      "step": 6860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004129305631492619,
+      "loss": 0.9116,
+      "step": 6870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004127938764352105,
+      "loss": 0.9569,
+      "step": 6880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004126571897211591,
+      "loss": 0.909,
+      "step": 6890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004125205030071077,
+      "loss": 0.814,
+      "step": 6900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004123838162930563,
+      "loss": 0.9581,
+      "step": 6910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041224712957900494,
+      "loss": 1.0221,
+      "step": 6920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004121104428649535,
+      "loss": 0.8015,
+      "step": 6930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041197375615090214,
+      "loss": 0.7999,
+      "step": 6940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004118370694368507,
+      "loss": 0.8941,
+      "step": 6950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041170038272279934,
+      "loss": 0.9369,
+      "step": 6960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004115636960087479,
+      "loss": 0.824,
+      "step": 6970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041142700929469654,
+      "loss": 0.9824,
+      "step": 6980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00041129032258064517,
+      "loss": 0.8795,
+      "step": 6990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004111536358665938,
+      "loss": 0.8961,
+      "step": 7000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004110169491525424,
+      "loss": 0.8199,
+      "step": 7010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000410880262438491,
+      "loss": 0.8908,
+      "step": 7020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004107435757244396,
+      "loss": 0.9221,
+      "step": 7030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004106068890103882,
+      "loss": 0.8772,
+      "step": 7040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004104702022963368,
+      "loss": 0.9297,
+      "step": 7050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00041033351558228545,
+      "loss": 0.9731,
+      "step": 7060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000410196828868234,
+      "loss": 0.9038,
+      "step": 7070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00041006014215418265,
+      "loss": 0.8985,
+      "step": 7080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004099234554401312,
+      "loss": 0.8589,
+      "step": 7090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040978676872607984,
+      "loss": 0.9197,
+      "step": 7100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0004096500820120284,
+      "loss": 0.93,
+      "step": 7110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00040951339529797704,
+      "loss": 0.9278,
+      "step": 7120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00040937670858392567,
+      "loss": 0.762,
+      "step": 7130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040924002186987424,
+      "loss": 0.8733,
+      "step": 7140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040910333515582287,
+      "loss": 0.8272,
+      "step": 7150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040896664844177144,
+      "loss": 1.008,
+      "step": 7160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040882996172772007,
+      "loss": 0.9744,
+      "step": 7170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00040869327501366864,
+      "loss": 0.8238,
+      "step": 7180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00040855658829961727,
+      "loss": 0.8728,
+      "step": 7190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0004084199015855659,
+      "loss": 0.8594,
+      "step": 7200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00040828321487151447,
+      "loss": 0.863,
+      "step": 7210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0004081465281574631,
+      "loss": 0.8859,
+      "step": 7220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00040800984144341167,
+      "loss": 0.9042,
+      "step": 7230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0004078731547293603,
+      "loss": 0.8675,
+      "step": 7240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0004077364680153089,
+      "loss": 0.8822,
+      "step": 7250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00040759978130125755,
+      "loss": 0.7771,
+      "step": 7260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0004074630945872062,
+      "loss": 0.9643,
+      "step": 7270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00040732640787315475,
+      "loss": 0.8555,
+      "step": 7280
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004071897211591034,
+      "loss": 0.8662,
+      "step": 7290
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00040705303444505195,
+      "loss": 0.7963,
+      "step": 7300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004069163477310006,
+      "loss": 0.8555,
+      "step": 7310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004067796610169492,
+      "loss": 0.9498,
+      "step": 7320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0004066429743028978,
+      "loss": 0.7529,
+      "step": 7330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0004065062875888464,
+      "loss": 0.8569,
+      "step": 7340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000406369600874795,
+      "loss": 0.8379,
+      "step": 7350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0004062329141607436,
+      "loss": 1.0122,
+      "step": 7360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004060962274466922,
+      "loss": 0.8528,
+      "step": 7370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004059595407326408,
+      "loss": 0.83,
+      "step": 7380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004058228540185894,
+      "loss": 0.8781,
+      "step": 7390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000405686167304538,
+      "loss": 0.8277,
+      "step": 7400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0004055494805904866,
+      "loss": 0.8536,
+      "step": 7410
+    },
+    {
+      "epoch": 2.0,
+      "eval_Bleu_1": 0.004787165221891267,
+      "eval_Bleu_2": 2.063194487625299e-11,
+      "eval_Bleu_3": 3.7314628962676555e-14,
+      "eval_Bleu_4": 1.6936871841068028e-15,
+      "eval_ROUGE_L": 0.0067366805770027495,
+      "eval_cer": 0.6323353293413174,
+      "eval_em": 0.3983855650522317,
+      "eval_f1": 0.5633121497389879,
+      "eval_loss": 0.935066819190979,
+      "eval_runtime": 316.1646,
+      "eval_samples_per_second": 13.322,
+      "eval_steps_per_second": 2.666,
+      "eval_wer": 0.7461864406779661,
+      "step": 7416
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0004054127938764352,
+      "loss": 0.7295,
+      "step": 7420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0004052761071623838,
+      "loss": 0.6611,
+      "step": 7430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0004051394204483324,
+      "loss": 0.6269,
+      "step": 7440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000405002733734281,
+      "loss": 0.6124,
+      "step": 7450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00040486604702022965,
+      "loss": 0.6172,
+      "step": 7460
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0004047293603061782,
+      "loss": 0.5909,
+      "step": 7470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00040459267359212685,
+      "loss": 0.5952,
+      "step": 7480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0004044559868780754,
+      "loss": 0.6686,
+      "step": 7490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00040431930016402405,
+      "loss": 0.6574,
+      "step": 7500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0004041826134499726,
+      "loss": 0.6852,
+      "step": 7510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00040404592673592125,
+      "loss": 0.6422,
+      "step": 7520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00040390924002186993,
+      "loss": 0.6566,
+      "step": 7530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0004037725533078185,
+      "loss": 0.6533,
+      "step": 7540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00040363586659376713,
+      "loss": 0.6349,
+      "step": 7550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0004034991798797157,
+      "loss": 0.6402,
+      "step": 7560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00040336249316566433,
+      "loss": 0.6156,
+      "step": 7570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0004032258064516129,
+      "loss": 0.6406,
+      "step": 7580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00040308911973756153,
+      "loss": 0.5764,
+      "step": 7590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00040295243302351016,
+      "loss": 0.6134,
+      "step": 7600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00040281574630945873,
+      "loss": 0.6604,
+      "step": 7610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00040267905959540736,
+      "loss": 0.6249,
+      "step": 7620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00040254237288135593,
+      "loss": 0.6281,
+      "step": 7630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00040240568616730456,
+      "loss": 0.571,
+      "step": 7640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00040226899945325313,
+      "loss": 0.6821,
+      "step": 7650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00040213231273920175,
+      "loss": 0.5187,
+      "step": 7660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0004019956260251504,
+      "loss": 0.7087,
+      "step": 7670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00040185893931109895,
+      "loss": 0.6407,
+      "step": 7680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0004017222525970476,
+      "loss": 0.6379,
+      "step": 7690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00040158556588299615,
+      "loss": 0.5982,
+      "step": 7700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0004014488791689448,
+      "loss": 0.6086,
+      "step": 7710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00040131219245489335,
+      "loss": 0.6376,
+      "step": 7720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000401175505740842,
+      "loss": 0.5145,
+      "step": 7730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0004010388190267906,
+      "loss": 0.6631,
+      "step": 7740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0004009021323127392,
+      "loss": 0.6705,
+      "step": 7750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0004007654455986878,
+      "loss": 0.5968,
+      "step": 7760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0004006287588846364,
+      "loss": 0.5618,
+      "step": 7770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000400492072170585,
+      "loss": 0.6502,
+      "step": 7780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0004003553854565337,
+      "loss": 0.5146,
+      "step": 7790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00040021869874248226,
+      "loss": 0.5598,
+      "step": 7800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0004000820120284309,
+      "loss": 0.6581,
+      "step": 7810
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039994532531437946,
+      "loss": 0.5982,
+      "step": 7820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003998086386003281,
+      "loss": 0.6699,
+      "step": 7830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039967195188627666,
+      "loss": 0.7018,
+      "step": 7840
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003995352651722253,
+      "loss": 0.6208,
+      "step": 7850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003993985784581739,
+      "loss": 0.7331,
+      "step": 7860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003992618917441225,
+      "loss": 0.7383,
+      "step": 7870
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003991252050300711,
+      "loss": 0.5751,
+      "step": 7880
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003989885183160197,
+      "loss": 0.6671,
+      "step": 7890
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003988518316019683,
+      "loss": 0.6164,
+      "step": 7900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003987151448879169,
+      "loss": 0.6223,
+      "step": 7910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003985784581738655,
+      "loss": 0.6022,
+      "step": 7920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00039844177145981414,
+      "loss": 0.6375,
+      "step": 7930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003983050847457627,
+      "loss": 0.6307,
+      "step": 7940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00039816839803171134,
+      "loss": 0.6844,
+      "step": 7950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003980317113176599,
+      "loss": 0.7049,
+      "step": 7960
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00039789502460360854,
+      "loss": 0.5999,
+      "step": 7970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003977583378895571,
+      "loss": 0.5818,
+      "step": 7980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00039762165117550574,
+      "loss": 0.6037,
+      "step": 7990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00039748496446145436,
+      "loss": 0.622,
+      "step": 8000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00039734827774740294,
+      "loss": 0.5557,
+      "step": 8010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00039721159103335156,
+      "loss": 0.6813,
+      "step": 8020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00039707490431930013,
+      "loss": 0.6181,
+      "step": 8030
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00039693821760524876,
+      "loss": 0.6672,
+      "step": 8040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00039680153089119733,
+      "loss": 0.5572,
+      "step": 8050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000396664844177146,
+      "loss": 0.6726,
+      "step": 8060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00039652815746309464,
+      "loss": 0.696,
+      "step": 8070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003963914707490432,
+      "loss": 0.6306,
+      "step": 8080
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00039625478403499184,
+      "loss": 0.6995,
+      "step": 8090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003961180973209404,
+      "loss": 0.5814,
+      "step": 8100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00039598141060688904,
+      "loss": 0.6546,
+      "step": 8110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003958447238928376,
+      "loss": 0.6117,
+      "step": 8120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00039570803717878624,
+      "loss": 0.6085,
+      "step": 8130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039557135046473487,
+      "loss": 0.6561,
+      "step": 8140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039543466375068344,
+      "loss": 0.6028,
+      "step": 8150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039529797703663207,
+      "loss": 0.6608,
+      "step": 8160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039516129032258064,
+      "loss": 0.568,
+      "step": 8170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00039502460360852927,
+      "loss": 0.5963,
+      "step": 8180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00039488791689447784,
+      "loss": 0.6255,
+      "step": 8190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00039475123018042647,
+      "loss": 0.6905,
+      "step": 8200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003946145434663751,
+      "loss": 0.6577,
+      "step": 8210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00039447785675232367,
+      "loss": 0.703,
+      "step": 8220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0003943411700382723,
+      "loss": 0.5881,
+      "step": 8230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00039420448332422086,
+      "loss": 0.7659,
+      "step": 8240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0003940677966101695,
+      "loss": 0.6576,
+      "step": 8250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00039393110989611806,
+      "loss": 0.6261,
+      "step": 8260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003937944231820667,
+      "loss": 0.6581,
+      "step": 8270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003936577364680153,
+      "loss": 0.665,
+      "step": 8280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003935210497539639,
+      "loss": 0.6103,
+      "step": 8290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003933843630399125,
+      "loss": 0.6095,
+      "step": 8300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003932476763258611,
+      "loss": 0.7314,
+      "step": 8310
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00039311098961180977,
+      "loss": 0.5826,
+      "step": 8320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0003929743028977584,
+      "loss": 0.7122,
+      "step": 8330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00039283761618370697,
+      "loss": 0.5759,
+      "step": 8340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0003927009294696556,
+      "loss": 0.7192,
+      "step": 8350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00039256424275560417,
+      "loss": 0.5979,
+      "step": 8360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003924275560415528,
+      "loss": 0.5847,
+      "step": 8370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00039229086932750137,
+      "loss": 0.6626,
+      "step": 8380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00039215418261345,
+      "loss": 0.6306,
+      "step": 8390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003920174958993986,
+      "loss": 0.5931,
+      "step": 8400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003918808091853472,
+      "loss": 0.6812,
+      "step": 8410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003917441224712958,
+      "loss": 0.5893,
+      "step": 8420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0003916074357572444,
+      "loss": 0.7556,
+      "step": 8430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000391470749043193,
+      "loss": 0.6927,
+      "step": 8440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003913340623291416,
+      "loss": 0.7186,
+      "step": 8450
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003911973756150902,
+      "loss": 0.6273,
+      "step": 8460
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00039106068890103885,
+      "loss": 0.6597,
+      "step": 8470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003909240021869874,
+      "loss": 0.6795,
+      "step": 8480
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00039078731547293605,
+      "loss": 0.7067,
+      "step": 8490
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003906506287588846,
+      "loss": 0.6724,
+      "step": 8500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00039051394204483325,
+      "loss": 0.6418,
+      "step": 8510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003903772553307818,
+      "loss": 0.6725,
+      "step": 8520
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00039024056861673045,
+      "loss": 0.6516,
+      "step": 8530
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003901038819026791,
+      "loss": 0.6679,
+      "step": 8540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00038996719518862765,
+      "loss": 0.5519,
+      "step": 8550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00038983050847457627,
+      "loss": 0.6568,
+      "step": 8560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00038969382176052485,
+      "loss": 0.6565,
+      "step": 8570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00038955713504647347,
+      "loss": 0.6417,
+      "step": 8580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003894204483324221,
+      "loss": 0.6166,
+      "step": 8590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003892837616183707,
+      "loss": 0.5645,
+      "step": 8600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00038914707490431935,
+      "loss": 0.6212,
+      "step": 8610
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003890103881902679,
+      "loss": 0.5845,
+      "step": 8620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00038887370147621655,
+      "loss": 0.5808,
+      "step": 8630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003887370147621651,
+      "loss": 0.5849,
+      "step": 8640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00038860032804811375,
+      "loss": 0.5757,
+      "step": 8650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003884636413340623,
+      "loss": 0.5872,
+      "step": 8660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00038832695462001095,
+      "loss": 0.697,
+      "step": 8670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003881902679059596,
+      "loss": 0.5864,
+      "step": 8680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00038805358119190815,
+      "loss": 0.6956,
+      "step": 8690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0003879168944778568,
+      "loss": 0.7193,
+      "step": 8700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00038778020776380535,
+      "loss": 0.6631,
+      "step": 8710
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000387643521049754,
+      "loss": 0.6087,
+      "step": 8720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00038750683433570255,
+      "loss": 0.5747,
+      "step": 8730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003873701476216512,
+      "loss": 0.6357,
+      "step": 8740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003872334609075998,
+      "loss": 0.6186,
+      "step": 8750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003870967741935484,
+      "loss": 0.6494,
+      "step": 8760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000386960087479497,
+      "loss": 0.6976,
+      "step": 8770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003868234007654456,
+      "loss": 0.6087,
+      "step": 8780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003866867140513942,
+      "loss": 0.5466,
+      "step": 8790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003865500273373428,
+      "loss": 0.5799,
+      "step": 8800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003864133406232914,
+      "loss": 0.6281,
+      "step": 8810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00038627665390924003,
+      "loss": 0.6624,
+      "step": 8820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003861399671951886,
+      "loss": 0.6389,
+      "step": 8830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00038600328048113723,
+      "loss": 0.6382,
+      "step": 8840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003858665937670858,
+      "loss": 0.6719,
+      "step": 8850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003857299070530345,
+      "loss": 0.5239,
+      "step": 8860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003855932203389831,
+      "loss": 0.6236,
+      "step": 8870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003854565336249317,
+      "loss": 0.6333,
+      "step": 8880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003853198469108803,
+      "loss": 0.66,
+      "step": 8890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003851831601968289,
+      "loss": 0.6734,
+      "step": 8900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003850464734827775,
+      "loss": 0.5506,
+      "step": 8910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003849097867687261,
+      "loss": 0.5672,
+      "step": 8920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003847731000546747,
+      "loss": 0.5532,
+      "step": 8930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00038463641334062333,
+      "loss": 0.6687,
+      "step": 8940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003844997266265719,
+      "loss": 0.7346,
+      "step": 8950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00038436303991252053,
+      "loss": 0.6483,
+      "step": 8960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003842263531984691,
+      "loss": 0.6783,
+      "step": 8970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00038408966648441773,
+      "loss": 0.6453,
+      "step": 8980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003839529797703663,
+      "loss": 0.6437,
+      "step": 8990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00038381629305631493,
+      "loss": 0.6082,
+      "step": 9000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00038367960634226356,
+      "loss": 0.7526,
+      "step": 9010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00038354291962821213,
+      "loss": 0.669,
+      "step": 9020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00038340623291416076,
+      "loss": 0.5517,
+      "step": 9030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00038326954620010933,
+      "loss": 0.7094,
+      "step": 9040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00038313285948605796,
+      "loss": 0.6256,
+      "step": 9050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00038299617277200653,
+      "loss": 0.6471,
+      "step": 9060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00038285948605795516,
+      "loss": 0.7337,
+      "step": 9070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0003827227993439038,
+      "loss": 0.6002,
+      "step": 9080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00038258611262985236,
+      "loss": 0.6053,
+      "step": 9090
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000382449425915801,
+      "loss": 0.6325,
+      "step": 9100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00038231273920174956,
+      "loss": 0.8111,
+      "step": 9110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00038217605248769824,
+      "loss": 0.6057,
+      "step": 9120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0003820393657736468,
+      "loss": 0.5804,
+      "step": 9130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00038190267905959544,
+      "loss": 0.6929,
+      "step": 9140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00038176599234554406,
+      "loss": 0.669,
+      "step": 9150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00038162930563149264,
+      "loss": 0.679,
+      "step": 9160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00038149261891744126,
+      "loss": 0.6397,
+      "step": 9170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00038135593220338984,
+      "loss": 0.5748,
+      "step": 9180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00038121924548933846,
+      "loss": 0.6388,
+      "step": 9190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00038108255877528704,
+      "loss": 0.6359,
+      "step": 9200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00038094587206123566,
+      "loss": 0.5492,
+      "step": 9210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003808091853471843,
+      "loss": 0.6681,
+      "step": 9220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00038067249863313286,
+      "loss": 0.6262,
+      "step": 9230
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003805358119190815,
+      "loss": 0.6138,
+      "step": 9240
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00038039912520503006,
+      "loss": 0.6696,
+      "step": 9250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003802624384909787,
+      "loss": 0.567,
+      "step": 9260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00038012575177692726,
+      "loss": 0.5801,
+      "step": 9270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003799890650628759,
+      "loss": 0.6522,
+      "step": 9280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003798523783488245,
+      "loss": 0.6927,
+      "step": 9290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003797156916347731,
+      "loss": 0.6886,
+      "step": 9300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003795790049207217,
+      "loss": 0.6504,
+      "step": 9310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003794423182066703,
+      "loss": 0.5569,
+      "step": 9320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003793056314926189,
+      "loss": 0.6761,
+      "step": 9330
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003791689447785675,
+      "loss": 0.6804,
+      "step": 9340
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003790322580645161,
+      "loss": 0.6274,
+      "step": 9350
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00037889557135046474,
+      "loss": 0.5387,
+      "step": 9360
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0003787588846364133,
+      "loss": 0.646,
+      "step": 9370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000378622197922362,
+      "loss": 0.619,
+      "step": 9380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00037848551120831057,
+      "loss": 0.7533,
+      "step": 9390
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003783488244942592,
+      "loss": 0.6357,
+      "step": 9400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003782121377802078,
+      "loss": 0.6246,
+      "step": 9410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003780754510661564,
+      "loss": 0.6893,
+      "step": 9420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000377938764352105,
+      "loss": 0.4835,
+      "step": 9430
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003778020776380536,
+      "loss": 0.6368,
+      "step": 9440
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003776653909240022,
+      "loss": 0.6741,
+      "step": 9450
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003775287042099508,
+      "loss": 0.6765,
+      "step": 9460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003773920174958994,
+      "loss": 0.568,
+      "step": 9470
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00037725533078184804,
+      "loss": 0.7014,
+      "step": 9480
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003771186440677966,
+      "loss": 0.6289,
+      "step": 9490
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00037698195735374524,
+      "loss": 0.6141,
+      "step": 9500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003768452706396938,
+      "loss": 0.7021,
+      "step": 9510
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00037670858392564244,
+      "loss": 0.6799,
+      "step": 9520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000376571897211591,
+      "loss": 0.5905,
+      "step": 9530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00037643521049753964,
+      "loss": 0.6543,
+      "step": 9540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00037629852378348827,
+      "loss": 0.5521,
+      "step": 9550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00037616183706943684,
+      "loss": 0.6335,
+      "step": 9560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00037602515035538547,
+      "loss": 0.6148,
+      "step": 9570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00037588846364133404,
+      "loss": 0.6573,
+      "step": 9580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00037575177692728267,
+      "loss": 0.6223,
+      "step": 9590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00037561509021323124,
+      "loss": 0.607,
+      "step": 9600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00037547840349917987,
+      "loss": 0.6802,
+      "step": 9610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0003753417167851285,
+      "loss": 0.6301,
+      "step": 9620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00037520503007107707,
+      "loss": 0.7136,
+      "step": 9630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003750683433570257,
+      "loss": 0.621,
+      "step": 9640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003749316566429743,
+      "loss": 0.6105,
+      "step": 9650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00037479496992892295,
+      "loss": 0.5834,
+      "step": 9660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003746582832148715,
+      "loss": 0.6373,
+      "step": 9670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00037452159650082015,
+      "loss": 0.6593,
+      "step": 9680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003743849097867688,
+      "loss": 0.6845,
+      "step": 9690
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00037424822307271735,
+      "loss": 0.5934,
+      "step": 9700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000374111536358666,
+      "loss": 0.6762,
+      "step": 9710
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00037397484964461455,
+      "loss": 0.7425,
+      "step": 9720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003738381629305632,
+      "loss": 0.581,
+      "step": 9730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00037370147621651175,
+      "loss": 0.6018,
+      "step": 9740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00037356478950246037,
+      "loss": 0.6767,
+      "step": 9750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000373428102788409,
+      "loss": 0.6048,
+      "step": 9760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00037329141607435757,
+      "loss": 0.6902,
+      "step": 9770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003731547293603062,
+      "loss": 0.7405,
+      "step": 9780
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00037301804264625477,
+      "loss": 0.6666,
+      "step": 9790
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003728813559322034,
+      "loss": 0.6536,
+      "step": 9800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00037274466921815197,
+      "loss": 0.5625,
+      "step": 9810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003726079825041006,
+      "loss": 0.604,
+      "step": 9820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003724712957900492,
+      "loss": 0.6173,
+      "step": 9830
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003723346090759978,
+      "loss": 0.6714,
+      "step": 9840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003721979223619464,
+      "loss": 0.6539,
+      "step": 9850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000372061235647895,
+      "loss": 0.567,
+      "step": 9860
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003719245489338436,
+      "loss": 0.5869,
+      "step": 9870
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003717878622197922,
+      "loss": 0.7388,
+      "step": 9880
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0003716511755057408,
+      "loss": 0.6649,
+      "step": 9890
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00037151448879168945,
+      "loss": 0.7055,
+      "step": 9900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000371377802077638,
+      "loss": 0.6132,
+      "step": 9910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003712411153635867,
+      "loss": 0.6872,
+      "step": 9920
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003711044286495353,
+      "loss": 0.6998,
+      "step": 9930
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003709677419354839,
+      "loss": 0.6299,
+      "step": 9940
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00037083105522143253,
+      "loss": 0.5275,
+      "step": 9950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0003706943685073811,
+      "loss": 0.6437,
+      "step": 9960
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00037055768179332973,
+      "loss": 0.6097,
+      "step": 9970
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0003704209950792783,
+      "loss": 0.5454,
+      "step": 9980
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00037028430836522693,
+      "loss": 0.6591,
+      "step": 9990
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003701476216511755,
+      "loss": 0.613,
+      "step": 10000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00037001093493712413,
+      "loss": 0.6697,
+      "step": 10010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00036987424822307276,
+      "loss": 0.6125,
+      "step": 10020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00036973756150902133,
+      "loss": 0.6238,
+      "step": 10030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00036960087479496995,
+      "loss": 0.6825,
+      "step": 10040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00036946418808091853,
+      "loss": 0.5865,
+      "step": 10050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00036932750136686715,
+      "loss": 0.7244,
+      "step": 10060
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0003691908146528157,
+      "loss": 0.6004,
+      "step": 10070
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00036905412793876435,
+      "loss": 0.6927,
+      "step": 10080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000368917441224713,
+      "loss": 0.604,
+      "step": 10090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00036878075451066155,
+      "loss": 0.5924,
+      "step": 10100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003686440677966102,
+      "loss": 0.6979,
+      "step": 10110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00036850738108255875,
+      "loss": 0.5898,
+      "step": 10120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003683706943685074,
+      "loss": 0.76,
+      "step": 10130
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00036823400765445595,
+      "loss": 0.651,
+      "step": 10140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003680973209404046,
+      "loss": 0.5123,
+      "step": 10150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003679606342263532,
+      "loss": 0.6671,
+      "step": 10160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003678239475123018,
+      "loss": 0.7331,
+      "step": 10170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036768726079825046,
+      "loss": 0.6894,
+      "step": 10180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036755057408419903,
+      "loss": 0.6118,
+      "step": 10190
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036741388737014766,
+      "loss": 0.6606,
+      "step": 10200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036727720065609623,
+      "loss": 0.5486,
+      "step": 10210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00036714051394204486,
+      "loss": 0.6415,
+      "step": 10220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003670038272279935,
+      "loss": 0.5394,
+      "step": 10230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00036686714051394206,
+      "loss": 0.7652,
+      "step": 10240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003667304537998907,
+      "loss": 0.612,
+      "step": 10250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00036659376708583926,
+      "loss": 0.5578,
+      "step": 10260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0003664570803717879,
+      "loss": 0.6377,
+      "step": 10270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00036632039365773646,
+      "loss": 0.6825,
+      "step": 10280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003661837069436851,
+      "loss": 0.6492,
+      "step": 10290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003660470202296337,
+      "loss": 0.649,
+      "step": 10300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003659103335155823,
+      "loss": 0.6659,
+      "step": 10310
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003657736468015309,
+      "loss": 0.6986,
+      "step": 10320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003656369600874795,
+      "loss": 0.7415,
+      "step": 10330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003655002733734281,
+      "loss": 0.6767,
+      "step": 10340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003653635866593767,
+      "loss": 0.6501,
+      "step": 10350
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003652268999453253,
+      "loss": 0.626,
+      "step": 10360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00036509021323127394,
+      "loss": 0.6423,
+      "step": 10370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0003649535265172225,
+      "loss": 0.6934,
+      "step": 10380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00036481683980317114,
+      "loss": 0.5858,
+      "step": 10390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0003646801530891197,
+      "loss": 0.5862,
+      "step": 10400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00036454346637506833,
+      "loss": 0.7088,
+      "step": 10410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003644067796610169,
+      "loss": 0.6334,
+      "step": 10420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00036427009294696553,
+      "loss": 0.5915,
+      "step": 10430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003641334062329142,
+      "loss": 0.6617,
+      "step": 10440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003639967195188628,
+      "loss": 0.6664,
+      "step": 10450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003638600328048114,
+      "loss": 0.6375,
+      "step": 10460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00036372334609076,
+      "loss": 0.6369,
+      "step": 10470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003635866593767086,
+      "loss": 0.6628,
+      "step": 10480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00036344997266265724,
+      "loss": 0.5654,
+      "step": 10490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003633132859486058,
+      "loss": 0.6234,
+      "step": 10500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00036317659923455444,
+      "loss": 0.6263,
+      "step": 10510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000363039912520503,
+      "loss": 0.5291,
+      "step": 10520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00036290322580645164,
+      "loss": 0.6604,
+      "step": 10530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003627665390924002,
+      "loss": 0.6365,
+      "step": 10540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00036262985237834884,
+      "loss": 0.5793,
+      "step": 10550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00036249316566429747,
+      "loss": 0.6084,
+      "step": 10560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00036235647895024604,
+      "loss": 0.8169,
+      "step": 10570
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00036221979223619467,
+      "loss": 0.6347,
+      "step": 10580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00036208310552214324,
+      "loss": 0.5688,
+      "step": 10590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00036194641880809186,
+      "loss": 0.7299,
+      "step": 10600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00036180973209404044,
+      "loss": 0.7025,
+      "step": 10610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00036167304537998906,
+      "loss": 0.6538,
+      "step": 10620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003615363586659377,
+      "loss": 0.7026,
+      "step": 10630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00036139967195188626,
+      "loss": 0.6557,
+      "step": 10640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003612629852378349,
+      "loss": 0.604,
+      "step": 10650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00036112629852378346,
+      "loss": 0.6603,
+      "step": 10660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003609896118097321,
+      "loss": 0.6522,
+      "step": 10670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00036085292509568066,
+      "loss": 0.657,
+      "step": 10680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003607162383816293,
+      "loss": 0.6425,
+      "step": 10690
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0003605795516675779,
+      "loss": 0.7497,
+      "step": 10700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00036044286495352654,
+      "loss": 0.5845,
+      "step": 10710
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00036030617823947517,
+      "loss": 0.6089,
+      "step": 10720
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00036016949152542374,
+      "loss": 0.6454,
+      "step": 10730
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00036003280481137237,
+      "loss": 0.5899,
+      "step": 10740
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00035989611809732094,
+      "loss": 0.6908,
+      "step": 10750
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00035975943138326957,
+      "loss": 0.5584,
+      "step": 10760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0003596227446692182,
+      "loss": 0.5891,
+      "step": 10770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00035948605795516677,
+      "loss": 0.5959,
+      "step": 10780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003593493712411154,
+      "loss": 0.6253,
+      "step": 10790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00035921268452706397,
+      "loss": 0.6839,
+      "step": 10800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003590759978130126,
+      "loss": 0.5751,
+      "step": 10810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00035893931109896117,
+      "loss": 0.6044,
+      "step": 10820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003588026243849098,
+      "loss": 0.6219,
+      "step": 10830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003586659376708584,
+      "loss": 0.7453,
+      "step": 10840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000358529250956807,
+      "loss": 0.605,
+      "step": 10850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003583925642427556,
+      "loss": 0.6156,
+      "step": 10860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003582558775287042,
+      "loss": 0.6251,
+      "step": 10870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003581191908146528,
+      "loss": 0.5345,
+      "step": 10880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0003579825041006014,
+      "loss": 0.7505,
+      "step": 10890
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00035784581738655,
+      "loss": 0.6425,
+      "step": 10900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00035770913067249865,
+      "loss": 0.6733,
+      "step": 10910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0003575724439584472,
+      "loss": 0.5932,
+      "step": 10920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00035743575724439585,
+      "loss": 0.6702,
+      "step": 10930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003572990705303444,
+      "loss": 0.5693,
+      "step": 10940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00035716238381629305,
+      "loss": 0.6836,
+      "step": 10950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003570256971022416,
+      "loss": 0.6918,
+      "step": 10960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00035688901038819024,
+      "loss": 0.6515,
+      "step": 10970
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003567523236741389,
+      "loss": 0.6472,
+      "step": 10980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003566156369600875,
+      "loss": 0.609,
+      "step": 10990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003564789502460361,
+      "loss": 0.6202,
+      "step": 11000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003563422635319847,
+      "loss": 0.582,
+      "step": 11010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003562055768179333,
+      "loss": 0.5269,
+      "step": 11020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00035606889010388195,
+      "loss": 0.7598,
+      "step": 11030
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003559322033898305,
+      "loss": 0.6242,
+      "step": 11040
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00035579551667577915,
+      "loss": 0.6669,
+      "step": 11050
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003556588299617277,
+      "loss": 0.6609,
+      "step": 11060
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00035552214324767635,
+      "loss": 0.6206,
+      "step": 11070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003553854565336249,
+      "loss": 0.6897,
+      "step": 11080
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00035524876981957355,
+      "loss": 0.5672,
+      "step": 11090
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003551120831055222,
+      "loss": 0.6783,
+      "step": 11100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00035497539639147075,
+      "loss": 0.6977,
+      "step": 11110
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003548387096774194,
+      "loss": 0.6765,
+      "step": 11120
+    },
+    {
+      "epoch": 3.0,
+      "eval_Bleu_1": 0.004397479653452061,
+      "eval_Bleu_2": 1.99724948145122e-11,
+      "eval_Bleu_3": 3.6792738476970824e-14,
+      "eval_Bleu_4": 1.686609611064112e-15,
+      "eval_ROUGE_L": 0.006335032725070497,
+      "eval_cer": 0.6010710286743586,
+      "eval_em": 0.42806267806267806,
+      "eval_f1": 0.5826430079528527,
+      "eval_loss": 0.8609268665313721,
+      "eval_runtime": 310.461,
+      "eval_samples_per_second": 13.567,
+      "eval_steps_per_second": 2.715,
+      "eval_wer": 0.710956416464891,
+      "step": 11124
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00035470202296336795,
+      "loss": 0.4318,
+      "step": 11130
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003545653362493166,
+      "loss": 0.4216,
+      "step": 11140
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00035442864953526515,
+      "loss": 0.4082,
+      "step": 11150
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003542919628212138,
+      "loss": 0.3968,
+      "step": 11160
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003541552761071624,
+      "loss": 0.4443,
+      "step": 11170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000354018589393111,
+      "loss": 0.4531,
+      "step": 11180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003538819026790596,
+      "loss": 0.4374,
+      "step": 11190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003537452159650082,
+      "loss": 0.4736,
+      "step": 11200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003536085292509568,
+      "loss": 0.4517,
+      "step": 11210
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003534718425369054,
+      "loss": 0.5002,
+      "step": 11220
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000353335155822854,
+      "loss": 0.3748,
+      "step": 11230
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003531984691088027,
+      "loss": 0.4636,
+      "step": 11240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00035306178239475125,
+      "loss": 0.4253,
+      "step": 11250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003529250956806999,
+      "loss": 0.3691,
+      "step": 11260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00035278840896664845,
+      "loss": 0.4742,
+      "step": 11270
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003526517222525971,
+      "loss": 0.4726,
+      "step": 11280
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00035251503553854565,
+      "loss": 0.4308,
+      "step": 11290
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003523783488244943,
+      "loss": 0.4734,
+      "step": 11300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003522416621104429,
+      "loss": 0.3985,
+      "step": 11310
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003521049753963915,
+      "loss": 0.4693,
+      "step": 11320
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003519682886823401,
+      "loss": 0.4363,
+      "step": 11330
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003518316019682887,
+      "loss": 0.4312,
+      "step": 11340
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003516949152542373,
+      "loss": 0.4221,
+      "step": 11350
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003515582285401859,
+      "loss": 0.4262,
+      "step": 11360
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003514215418261345,
+      "loss": 0.5569,
+      "step": 11370
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00035128485511208313,
+      "loss": 0.3885,
+      "step": 11380
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003511481683980317,
+      "loss": 0.4868,
+      "step": 11390
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00035101148168398033,
+      "loss": 0.4962,
+      "step": 11400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003508747949699289,
+      "loss": 0.4922,
+      "step": 11410
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00035073810825587753,
+      "loss": 0.4106,
+      "step": 11420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003506014215418261,
+      "loss": 0.4448,
+      "step": 11430
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00035046473482777473,
+      "loss": 0.4448,
+      "step": 11440
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00035032804811372336,
+      "loss": 0.4259,
+      "step": 11450
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00035019136139967193,
+      "loss": 0.4661,
+      "step": 11460
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00035005467468562056,
+      "loss": 0.4195,
+      "step": 11470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00034991798797156913,
+      "loss": 0.3741,
+      "step": 11480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00034978130125751776,
+      "loss": 0.4713,
+      "step": 11490
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00034964461454346633,
+      "loss": 0.3759,
+      "step": 11500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000349507927829415,
+      "loss": 0.4911,
+      "step": 11510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00034937124111536364,
+      "loss": 0.3787,
+      "step": 11520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003492345544013122,
+      "loss": 0.4303,
+      "step": 11530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00034909786768726084,
+      "loss": 0.416,
+      "step": 11540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003489611809732094,
+      "loss": 0.4603,
+      "step": 11550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00034882449425915804,
+      "loss": 0.3919,
+      "step": 11560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00034868780754510666,
+      "loss": 0.4392,
+      "step": 11570
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00034855112083105523,
+      "loss": 0.468,
+      "step": 11580
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00034841443411700386,
+      "loss": 0.4793,
+      "step": 11590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00034827774740295243,
+      "loss": 0.3882,
+      "step": 11600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00034814106068890106,
+      "loss": 0.4438,
+      "step": 11610
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00034800437397484963,
+      "loss": 0.464,
+      "step": 11620
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00034786768726079826,
+      "loss": 0.4044,
+      "step": 11630
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003477310005467469,
+      "loss": 0.3931,
+      "step": 11640
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00034759431383269546,
+      "loss": 0.4202,
+      "step": 11650
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003474576271186441,
+      "loss": 0.3688,
+      "step": 11660
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00034732094040459266,
+      "loss": 0.4396,
+      "step": 11670
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003471842536905413,
+      "loss": 0.416,
+      "step": 11680
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00034704756697648986,
+      "loss": 0.4303,
+      "step": 11690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003469108802624385,
+      "loss": 0.3648,
+      "step": 11700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003467741935483871,
+      "loss": 0.4371,
+      "step": 11710
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003466375068343357,
+      "loss": 0.3831,
+      "step": 11720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003465008201202843,
+      "loss": 0.4227,
+      "step": 11730
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003463641334062329,
+      "loss": 0.396,
+      "step": 11740
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003462274466921815,
+      "loss": 0.367,
+      "step": 11750
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003460907599781301,
+      "loss": 0.4914,
+      "step": 11760
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00034595407326407877,
+      "loss": 0.4927,
+      "step": 11770
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003458173865500274,
+      "loss": 0.4074,
+      "step": 11780
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00034568069983597596,
+      "loss": 0.4713,
+      "step": 11790
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003455440131219246,
+      "loss": 0.4667,
+      "step": 11800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00034540732640787316,
+      "loss": 0.5158,
+      "step": 11810
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0003452706396938218,
+      "loss": 0.4317,
+      "step": 11820
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00034513395297977036,
+      "loss": 0.4616,
+      "step": 11830
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000344997266265719,
+      "loss": 0.3914,
+      "step": 11840
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0003448605795516676,
+      "loss": 0.5355,
+      "step": 11850
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0003447238928376162,
+      "loss": 0.5247,
+      "step": 11860
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0003445872061235648,
+      "loss": 0.4613,
+      "step": 11870
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0003444505194095134,
+      "loss": 0.4709,
+      "step": 11880
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000344313832695462,
+      "loss": 0.4436,
+      "step": 11890
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0003441771459814106,
+      "loss": 0.4435,
+      "step": 11900
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0003440404592673592,
+      "loss": 0.4407,
+      "step": 11910
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00034390377255330784,
+      "loss": 0.4147,
+      "step": 11920
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0003437670858392564,
+      "loss": 0.3545,
+      "step": 11930
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00034363039912520504,
+      "loss": 0.4475,
+      "step": 11940
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0003434937124111536,
+      "loss": 0.4839,
+      "step": 11950
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00034335702569710224,
+      "loss": 0.4505,
+      "step": 11960
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003432203389830508,
+      "loss": 0.4312,
+      "step": 11970
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00034308365226899944,
+      "loss": 0.4647,
+      "step": 11980
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00034294696555494807,
+      "loss": 0.4396,
+      "step": 11990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034281027884089664,
+      "loss": 0.4539,
+      "step": 12000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034267359212684527,
+      "loss": 0.4807,
+      "step": 12010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034253690541279384,
+      "loss": 0.4486,
+      "step": 12020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034240021869874247,
+      "loss": 0.3994,
+      "step": 12030
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003422635319846911,
+      "loss": 0.4302,
+      "step": 12040
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003421268452706397,
+      "loss": 0.5099,
+      "step": 12050
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00034199015855658835,
+      "loss": 0.4733,
+      "step": 12060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0003418534718425369,
+      "loss": 0.4948,
+      "step": 12070
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00034171678512848555,
+      "loss": 0.4561,
+      "step": 12080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0003415800984144341,
+      "loss": 0.4272,
+      "step": 12090
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00034144341170038275,
+      "loss": 0.4888,
+      "step": 12100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0003413067249863314,
+      "loss": 0.4527,
+      "step": 12110
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00034117003827227995,
+      "loss": 0.5052,
+      "step": 12120
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00034103335155822857,
+      "loss": 0.4486,
+      "step": 12130
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00034089666484417715,
+      "loss": 0.4159,
+      "step": 12140
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00034075997813012577,
+      "loss": 0.4071,
+      "step": 12150
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00034062329141607434,
+      "loss": 0.4706,
+      "step": 12160
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00034048660470202297,
+      "loss": 0.5106,
+      "step": 12170
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0003403499179879716,
+      "loss": 0.4432,
+      "step": 12180
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00034021323127392017,
+      "loss": 0.4138,
+      "step": 12190
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0003400765445598688,
+      "loss": 0.4508,
+      "step": 12200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00033993985784581737,
+      "loss": 0.4927,
+      "step": 12210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000339803171131766,
+      "loss": 0.568,
+      "step": 12220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00033966648441771457,
+      "loss": 0.4575,
+      "step": 12230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0003395297977036632,
+      "loss": 0.3961,
+      "step": 12240
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0003393931109896118,
+      "loss": 0.5414,
+      "step": 12250
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0003392564242755604,
+      "loss": 0.4036,
+      "step": 12260
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000339119737561509,
+      "loss": 0.4214,
+      "step": 12270
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0003389830508474576,
+      "loss": 0.4896,
+      "step": 12280
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0003388463641334062,
+      "loss": 0.5003,
+      "step": 12290
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00033870967741935485,
+      "loss": 0.4974,
+      "step": 12300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0003385729907053035,
+      "loss": 0.4322,
+      "step": 12310
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0003384363039912521,
+      "loss": 0.4649,
+      "step": 12320
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003382996172772007,
+      "loss": 0.4708,
+      "step": 12330
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003381629305631493,
+      "loss": 0.4508,
+      "step": 12340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003380262438490979,
+      "loss": 0.4625,
+      "step": 12350
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003378895571350465,
+      "loss": 0.4972,
+      "step": 12360
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0003377528704209951,
+      "loss": 0.5337,
+      "step": 12370
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0003376161837069437,
+      "loss": 0.4726,
+      "step": 12380
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00033747949699289233,
+      "loss": 0.4875,
+      "step": 12390
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0003373428102788409,
+      "loss": 0.4635,
+      "step": 12400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00033720612356478953,
+      "loss": 0.4579,
+      "step": 12410
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0003370694368507381,
+      "loss": 0.4485,
+      "step": 12420
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0003369327501366867,
+      "loss": 0.5124,
+      "step": 12430
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0003367960634226353,
+      "loss": 0.5158,
+      "step": 12440
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0003366593767085839,
+      "loss": 0.4967,
+      "step": 12450
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00033652268999453255,
+      "loss": 0.4264,
+      "step": 12460
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0003363860032804811,
+      "loss": 0.4205,
+      "step": 12470
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00033624931656642975,
+      "loss": 0.4341,
+      "step": 12480
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0003361126298523783,
+      "loss": 0.4261,
+      "step": 12490
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00033597594313832695,
+      "loss": 0.5107,
+      "step": 12500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0003358392564242755,
+      "loss": 0.5024,
+      "step": 12510
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033570256971022415,
+      "loss": 0.4824,
+      "step": 12520
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0003355658829961728,
+      "loss": 0.5004,
+      "step": 12530
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033542919628212135,
+      "loss": 0.4083,
+      "step": 12540
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033529250956807,
+      "loss": 0.4967,
+      "step": 12550
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00033515582285401855,
+      "loss": 0.3896,
+      "step": 12560
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00033501913613996723,
+      "loss": 0.5225,
+      "step": 12570
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0003348824494259158,
+      "loss": 0.3796,
+      "step": 12580
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033474576271186443,
+      "loss": 0.4531,
+      "step": 12590
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033460907599781306,
+      "loss": 0.3716,
+      "step": 12600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033447238928376163,
+      "loss": 0.4515,
+      "step": 12610
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033433570256971026,
+      "loss": 0.4373,
+      "step": 12620
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00033419901585565883,
+      "loss": 0.4723,
+      "step": 12630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00033406232914160746,
+      "loss": 0.5386,
+      "step": 12640
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0003339256424275561,
+      "loss": 0.5038,
+      "step": 12650
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00033378895571350466,
+      "loss": 0.4959,
+      "step": 12660
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0003336522689994533,
+      "loss": 0.3938,
+      "step": 12670
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00033351558228540186,
+      "loss": 0.4206,
+      "step": 12680
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0003333788955713505,
+      "loss": 0.4329,
+      "step": 12690
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00033324220885729906,
+      "loss": 0.3984,
+      "step": 12700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003331055221432477,
+      "loss": 0.4045,
+      "step": 12710
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003329688354291963,
+      "loss": 0.5129,
+      "step": 12720
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003328321487151449,
+      "loss": 0.5328,
+      "step": 12730
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0003326954620010935,
+      "loss": 0.4995,
+      "step": 12740
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0003325587752870421,
+      "loss": 0.4975,
+      "step": 12750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0003324220885729907,
+      "loss": 0.5148,
+      "step": 12760
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0003322854018589393,
+      "loss": 0.4106,
+      "step": 12770
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0003321487151448879,
+      "loss": 0.5092,
+      "step": 12780
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00033201202843083653,
+      "loss": 0.4983,
+      "step": 12790
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0003318753417167851,
+      "loss": 0.4226,
+      "step": 12800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00033173865500273373,
+      "loss": 0.481,
+      "step": 12810
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0003316019682886823,
+      "loss": 0.5246,
+      "step": 12820
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.000331465281574631,
+      "loss": 0.5375,
+      "step": 12830
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00033132859486057956,
+      "loss": 0.4375,
+      "step": 12840
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0003311919081465282,
+      "loss": 0.5102,
+      "step": 12850
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0003310552214324768,
+      "loss": 0.5292,
+      "step": 12860
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0003309185347184254,
+      "loss": 0.4279,
+      "step": 12870
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000330781848004374,
+      "loss": 0.4835,
+      "step": 12880
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003306451612903226,
+      "loss": 0.5102,
+      "step": 12890
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003305084745762712,
+      "loss": 0.5117,
+      "step": 12900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003303717878622198,
+      "loss": 0.4562,
+      "step": 12910
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003302351011481684,
+      "loss": 0.4323,
+      "step": 12920
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00033009841443411704,
+      "loss": 0.4413,
+      "step": 12930
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0003299617277200656,
+      "loss": 0.4483,
+      "step": 12940
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00032982504100601424,
+      "loss": 0.4993,
+      "step": 12950
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0003296883542919628,
+      "loss": 0.4878,
+      "step": 12960
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00032955166757791144,
+      "loss": 0.4745,
+      "step": 12970
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00032941498086386,
+      "loss": 0.4717,
+      "step": 12980
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00032927829414980864,
+      "loss": 0.4304,
+      "step": 12990
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00032914160743575726,
+      "loss": 0.3912,
+      "step": 13000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00032900492072170584,
+      "loss": 0.461,
+      "step": 13010
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00032886823400765446,
+      "loss": 0.4735,
+      "step": 13020
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00032873154729360304,
+      "loss": 0.4021,
+      "step": 13030
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00032859486057955166,
+      "loss": 0.4907,
+      "step": 13040
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00032845817386550024,
+      "loss": 0.4431,
+      "step": 13050
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00032832148715144886,
+      "loss": 0.4676,
+      "step": 13060
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0003281848004373975,
+      "loss": 0.4964,
+      "step": 13070
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00032804811372334606,
+      "loss": 0.4978,
+      "step": 13080
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00032791142700929474,
+      "loss": 0.4765,
+      "step": 13090
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0003277747402952433,
+      "loss": 0.5137,
+      "step": 13100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00032763805358119194,
+      "loss": 0.4616,
+      "step": 13110
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0003275013668671405,
+      "loss": 0.4381,
+      "step": 13120
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00032736468015308914,
+      "loss": 0.4951,
+      "step": 13130
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00032722799343903777,
+      "loss": 0.4292,
+      "step": 13140
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00032709130672498634,
+      "loss": 0.4743,
+      "step": 13150
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00032695462001093497,
+      "loss": 0.4632,
+      "step": 13160
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00032681793329688354,
+      "loss": 0.4908,
+      "step": 13170
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00032668124658283217,
+      "loss": 0.5342,
+      "step": 13180
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0003265445598687808,
+      "loss": 0.4318,
+      "step": 13190
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00032640787315472937,
+      "loss": 0.5503,
+      "step": 13200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.000326271186440678,
+      "loss": 0.481,
+      "step": 13210
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00032613449972662657,
+      "loss": 0.4732,
+      "step": 13220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003259978130125752,
+      "loss": 0.5207,
+      "step": 13230
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00032586112629852377,
+      "loss": 0.5003,
+      "step": 13240
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003257244395844724,
+      "loss": 0.4385,
+      "step": 13250
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000325587752870421,
+      "loss": 0.54,
+      "step": 13260
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003254510661563696,
+      "loss": 0.5051,
+      "step": 13270
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003253143794423182,
+      "loss": 0.4611,
+      "step": 13280
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003251776927282668,
+      "loss": 0.4909,
+      "step": 13290
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003250410060142154,
+      "loss": 0.4014,
+      "step": 13300
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000324904319300164,
+      "loss": 0.5709,
+      "step": 13310
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003247676325861126,
+      "loss": 0.4838,
+      "step": 13320
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00032463094587206125,
+      "loss": 0.3833,
+      "step": 13330
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003244942591580098,
+      "loss": 0.519,
+      "step": 13340
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00032435757244395844,
+      "loss": 0.4491,
+      "step": 13350
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00032422088572990707,
+      "loss": 0.4203,
+      "step": 13360
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003240841990158557,
+      "loss": 0.4278,
+      "step": 13370
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00032394751230180427,
+      "loss": 0.3969,
+      "step": 13380
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003238108255877529,
+      "loss": 0.4296,
+      "step": 13390
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003236741388737015,
+      "loss": 0.5414,
+      "step": 13400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003235374521596501,
+      "loss": 0.425,
+      "step": 13410
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003234007654455987,
+      "loss": 0.4577,
+      "step": 13420
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003232640787315473,
+      "loss": 0.4829,
+      "step": 13430
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003231273920174959,
+      "loss": 0.4663,
+      "step": 13440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003229907053034445,
+      "loss": 0.4648,
+      "step": 13450
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003228540185893931,
+      "loss": 0.4911,
+      "step": 13460
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00032271733187534175,
+      "loss": 0.552,
+      "step": 13470
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003225806451612903,
+      "loss": 0.4521,
+      "step": 13480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00032244395844723895,
+      "loss": 0.4491,
+      "step": 13490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003223072717331875,
+      "loss": 0.492,
+      "step": 13500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00032217058501913615,
+      "loss": 0.5349,
+      "step": 13510
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003220338983050847,
+      "loss": 0.4521,
+      "step": 13520
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00032189721159103335,
+      "loss": 0.482,
+      "step": 13530
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.000321760524876982,
+      "loss": 0.4166,
+      "step": 13540
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00032162383816293055,
+      "loss": 0.4659,
+      "step": 13550
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003214871514488792,
+      "loss": 0.4303,
+      "step": 13560
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00032135046473482775,
+      "loss": 0.4585,
+      "step": 13570
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003212137780207764,
+      "loss": 0.4951,
+      "step": 13580
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00032107709130672495,
+      "loss": 0.47,
+      "step": 13590
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003209404045926736,
+      "loss": 0.4824,
+      "step": 13600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003208037178786222,
+      "loss": 0.4438,
+      "step": 13610
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00032066703116457077,
+      "loss": 0.4699,
+      "step": 13620
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00032053034445051945,
+      "loss": 0.501,
+      "step": 13630
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000320393657736468,
+      "loss": 0.4582,
+      "step": 13640
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00032025697102241665,
+      "loss": 0.5212,
+      "step": 13650
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003201202843083652,
+      "loss": 0.3686,
+      "step": 13660
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00031998359759431385,
+      "loss": 0.5904,
+      "step": 13670
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003198469108802625,
+      "loss": 0.4132,
+      "step": 13680
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00031971022416621105,
+      "loss": 0.4636,
+      "step": 13690
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003195735374521597,
+      "loss": 0.4957,
+      "step": 13700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00031943685073810825,
+      "loss": 0.4818,
+      "step": 13710
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003193001640240569,
+      "loss": 0.4854,
+      "step": 13720
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003191634773100055,
+      "loss": 0.445,
+      "step": 13730
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003190267905959541,
+      "loss": 0.3772,
+      "step": 13740
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003188901038819027,
+      "loss": 0.4502,
+      "step": 13750
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003187534171678513,
+      "loss": 0.5602,
+      "step": 13760
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003186167304537999,
+      "loss": 0.5244,
+      "step": 13770
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003184800437397485,
+      "loss": 0.418,
+      "step": 13780
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003183433570256971,
+      "loss": 0.4476,
+      "step": 13790
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00031820667031164573,
+      "loss": 0.4515,
+      "step": 13800
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003180699835975943,
+      "loss": 0.4072,
+      "step": 13810
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031793329688354293,
+      "loss": 0.4893,
+      "step": 13820
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003177966101694915,
+      "loss": 0.5908,
+      "step": 13830
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031765992345544013,
+      "loss": 0.4885,
+      "step": 13840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003175232367413887,
+      "loss": 0.4584,
+      "step": 13850
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031738655002733733,
+      "loss": 0.4478,
+      "step": 13860
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031724986331328596,
+      "loss": 0.5106,
+      "step": 13870
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031711317659923453,
+      "loss": 0.4812,
+      "step": 13880
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003169764898851832,
+      "loss": 0.589,
+      "step": 13890
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003168398031711318,
+      "loss": 0.3986,
+      "step": 13900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003167031164570804,
+      "loss": 0.4529,
+      "step": 13910
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.000316566429743029,
+      "loss": 0.4816,
+      "step": 13920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003164297430289776,
+      "loss": 0.4993,
+      "step": 13930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00031629305631492624,
+      "loss": 0.4454,
+      "step": 13940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003161563696008748,
+      "loss": 0.4247,
+      "step": 13950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00031601968288682343,
+      "loss": 0.5223,
+      "step": 13960
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000315882996172772,
+      "loss": 0.4561,
+      "step": 13970
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00031574630945872063,
+      "loss": 0.5129,
+      "step": 13980
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003156096227446692,
+      "loss": 0.4787,
+      "step": 13990
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00031547293603061783,
+      "loss": 0.4961,
+      "step": 14000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00031533624931656646,
+      "loss": 0.503,
+      "step": 14010
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00031519956260251503,
+      "loss": 0.5662,
+      "step": 14020
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00031506287588846366,
+      "loss": 0.5464,
+      "step": 14030
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031492618917441223,
+      "loss": 0.403,
+      "step": 14040
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031478950246036086,
+      "loss": 0.4859,
+      "step": 14050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031465281574630943,
+      "loss": 0.5636,
+      "step": 14060
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031451612903225806,
+      "loss": 0.5513,
+      "step": 14070
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003143794423182067,
+      "loss": 0.4626,
+      "step": 14080
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00031424275560415526,
+      "loss": 0.5501,
+      "step": 14090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003141060688901039,
+      "loss": 0.4445,
+      "step": 14100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00031396938217605246,
+      "loss": 0.4572,
+      "step": 14110
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0003138326954620011,
+      "loss": 0.5867,
+      "step": 14120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00031369600874794966,
+      "loss": 0.4919,
+      "step": 14130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0003135593220338983,
+      "loss": 0.5087,
+      "step": 14140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00031342263531984697,
+      "loss": 0.4141,
+      "step": 14150
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00031328594860579554,
+      "loss": 0.5692,
+      "step": 14160
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00031314926189174416,
+      "loss": 0.4468,
+      "step": 14170
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00031301257517769274,
+      "loss": 0.4027,
+      "step": 14180
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00031287588846364136,
+      "loss": 0.4867,
+      "step": 14190
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00031273920174958994,
+      "loss": 0.4623,
+      "step": 14200
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00031260251503553856,
+      "loss": 0.4266,
+      "step": 14210
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0003124658283214872,
+      "loss": 0.4448,
+      "step": 14220
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00031232914160743576,
+      "loss": 0.3705,
+      "step": 14230
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0003121924548933844,
+      "loss": 0.5618,
+      "step": 14240
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00031205576817933296,
+      "loss": 0.5412,
+      "step": 14250
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0003119190814652816,
+      "loss": 0.541,
+      "step": 14260
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0003117823947512302,
+      "loss": 0.4538,
+      "step": 14270
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0003116457080371788,
+      "loss": 0.4901,
+      "step": 14280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0003115090213231274,
+      "loss": 0.4929,
+      "step": 14290
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.000311372334609076,
+      "loss": 0.4703,
+      "step": 14300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003112356478950246,
+      "loss": 0.4663,
+      "step": 14310
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003110989611809732,
+      "loss": 0.4455,
+      "step": 14320
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003109622744669218,
+      "loss": 0.4818,
+      "step": 14330
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00031082558775287044,
+      "loss": 0.4551,
+      "step": 14340
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000310688901038819,
+      "loss": 0.5484,
+      "step": 14350
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00031055221432476764,
+      "loss": 0.4363,
+      "step": 14360
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0003104155276107162,
+      "loss": 0.4473,
+      "step": 14370
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00031027884089666484,
+      "loss": 0.4501,
+      "step": 14380
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0003101421541826134,
+      "loss": 0.4089,
+      "step": 14390
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00031000546746856204,
+      "loss": 0.4856,
+      "step": 14400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00030986878075451067,
+      "loss": 0.4573,
+      "step": 14410
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003097320940404593,
+      "loss": 0.5404,
+      "step": 14420
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003095954073264079,
+      "loss": 0.4865,
+      "step": 14430
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003094587206123565,
+      "loss": 0.4035,
+      "step": 14440
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003093220338983051,
+      "loss": 0.3731,
+      "step": 14450
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003091853471842537,
+      "loss": 0.5303,
+      "step": 14460
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003090486604702023,
+      "loss": 0.514,
+      "step": 14470
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00030891197375615095,
+      "loss": 0.4666,
+      "step": 14480
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0003087752870420995,
+      "loss": 0.4477,
+      "step": 14490
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00030863860032804815,
+      "loss": 0.464,
+      "step": 14500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0003085019136139967,
+      "loss": 0.4559,
+      "step": 14510
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00030836522689994534,
+      "loss": 0.4817,
+      "step": 14520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0003082285401858939,
+      "loss": 0.4081,
+      "step": 14530
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00030809185347184254,
+      "loss": 0.5085,
+      "step": 14540
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00030795516675779117,
+      "loss": 0.5237,
+      "step": 14550
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030781848004373974,
+      "loss": 0.5519,
+      "step": 14560
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030768179332968837,
+      "loss": 0.5082,
+      "step": 14570
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030754510661563694,
+      "loss": 0.5188,
+      "step": 14580
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030740841990158557,
+      "loss": 0.5127,
+      "step": 14590
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00030727173318753414,
+      "loss": 0.4359,
+      "step": 14600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00030713504647348277,
+      "loss": 0.4426,
+      "step": 14610
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0003069983597594314,
+      "loss": 0.4267,
+      "step": 14620
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00030686167304537997,
+      "loss": 0.4744,
+      "step": 14630
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0003067249863313286,
+      "loss": 0.4821,
+      "step": 14640
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00030658829961727717,
+      "loss": 0.4533,
+      "step": 14650
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0003064516129032258,
+      "loss": 0.4245,
+      "step": 14660
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00030631492618917437,
+      "loss": 0.5778,
+      "step": 14670
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.000306178239475123,
+      "loss": 0.4765,
+      "step": 14680
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0003060415527610717,
+      "loss": 0.4328,
+      "step": 14690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00030590486604702025,
+      "loss": 0.5323,
+      "step": 14700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0003057681793329689,
+      "loss": 0.5166,
+      "step": 14710
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00030563149261891745,
+      "loss": 0.4632,
+      "step": 14720
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0003054948059048661,
+      "loss": 0.468,
+      "step": 14730
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00030535811919081465,
+      "loss": 0.4356,
+      "step": 14740
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0003052214324767633,
+      "loss": 0.5089,
+      "step": 14750
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0003050847457627119,
+      "loss": 0.4713,
+      "step": 14760
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0003049480590486605,
+      "loss": 0.4542,
+      "step": 14770
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003048113723346091,
+      "loss": 0.4563,
+      "step": 14780
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003046746856205577,
+      "loss": 0.4189,
+      "step": 14790
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003045379989065063,
+      "loss": 0.5251,
+      "step": 14800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003044013121924549,
+      "loss": 0.4438,
+      "step": 14810
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0003042646254784035,
+      "loss": 0.4691,
+      "step": 14820
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0003041279387643521,
+      "loss": 0.4742,
+      "step": 14830
+    },
+    {
+      "epoch": 4.0,
+      "eval_Bleu_1": 0.005250913924891642,
+      "eval_Bleu_2": 2.2016236308922055e-11,
+      "eval_Bleu_3": 3.9498209106053914e-14,
+      "eval_Bleu_4": 1.788017365981181e-15,
+      "eval_ROUGE_L": 0.007519510418447302,
+      "eval_cer": 0.6017039092546614,
+      "eval_em": 0.4472934472934473,
+      "eval_f1": 0.5993383811283113,
+      "eval_loss": 0.8578809499740601,
+      "eval_runtime": 310.0126,
+      "eval_samples_per_second": 13.587,
+      "eval_steps_per_second": 2.719,
+      "eval_wer": 0.7047820823244552,
+      "step": 14832
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0003039912520503007,
+      "loss": 0.3526,
+      "step": 14840
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0003038545653362493,
+      "loss": 0.3038,
+      "step": 14850
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003037178786221979,
+      "loss": 0.3271,
+      "step": 14860
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0003035811919081465,
+      "loss": 0.2674,
+      "step": 14870
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00030344450519409515,
+      "loss": 0.3066,
+      "step": 14880
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003033078184800437,
+      "loss": 0.3313,
+      "step": 14890
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00030317113176599235,
+      "loss": 0.3846,
+      "step": 14900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003030344450519409,
+      "loss": 0.3788,
+      "step": 14910
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00030289775833788955,
+      "loss": 0.3105,
+      "step": 14920
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003027610716238381,
+      "loss": 0.2872,
+      "step": 14930
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00030262438490978675,
+      "loss": 0.3016,
+      "step": 14940
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00030248769819573543,
+      "loss": 0.2949,
+      "step": 14950
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.000302351011481684,
+      "loss": 0.314,
+      "step": 14960
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00030221432476763263,
+      "loss": 0.3031,
+      "step": 14970
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003020776380535812,
+      "loss": 0.393,
+      "step": 14980
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00030194095133952983,
+      "loss": 0.3249,
+      "step": 14990
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003018042646254784,
+      "loss": 0.2976,
+      "step": 15000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00030166757791142703,
+      "loss": 0.3093,
+      "step": 15010
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00030153089119737566,
+      "loss": 0.2768,
+      "step": 15020
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00030139420448332423,
+      "loss": 0.2975,
+      "step": 15030
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00030125751776927286,
+      "loss": 0.2719,
+      "step": 15040
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00030112083105522143,
+      "loss": 0.3252,
+      "step": 15050
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00030098414434117006,
+      "loss": 0.3193,
+      "step": 15060
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00030084745762711863,
+      "loss": 0.3179,
+      "step": 15070
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00030071077091306726,
+      "loss": 0.3641,
+      "step": 15080
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003005740841990159,
+      "loss": 0.3591,
+      "step": 15090
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00030043739748496445,
+      "loss": 0.2555,
+      "step": 15100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003003007107709131,
+      "loss": 0.3096,
+      "step": 15110
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00030016402405686165,
+      "loss": 0.3403,
+      "step": 15120
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003000273373428103,
+      "loss": 0.3508,
+      "step": 15130
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00029989065062875885,
+      "loss": 0.3658,
+      "step": 15140
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0002997539639147075,
+      "loss": 0.327,
+      "step": 15150
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0002996172772006561,
+      "loss": 0.3064,
+      "step": 15160
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0002994805904866047,
+      "loss": 0.3019,
+      "step": 15170
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0002993439037725533,
+      "loss": 0.3421,
+      "step": 15180
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002992072170585019,
+      "loss": 0.3208,
+      "step": 15190
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002990705303444505,
+      "loss": 0.3234,
+      "step": 15200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002989338436303991,
+      "loss": 0.342,
+      "step": 15210
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00029879715691634776,
+      "loss": 0.3874,
+      "step": 15220
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0002986604702022964,
+      "loss": 0.274,
+      "step": 15230
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00029852378348824496,
+      "loss": 0.2899,
+      "step": 15240
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0002983870967741936,
+      "loss": 0.3204,
+      "step": 15250
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00029825041006014216,
+      "loss": 0.3303,
+      "step": 15260
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0002981137233460908,
+      "loss": 0.3221,
+      "step": 15270
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00029797703663203936,
+      "loss": 0.2668,
+      "step": 15280
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000297840349917988,
+      "loss": 0.3058,
+      "step": 15290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002977036632039366,
+      "loss": 0.3072,
+      "step": 15300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002975669764898852,
+      "loss": 0.3136,
+      "step": 15310
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002974302897758338,
+      "loss": 0.3563,
+      "step": 15320
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002972936030617824,
+      "loss": 0.3212,
+      "step": 15330
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.000297156916347731,
+      "loss": 0.3192,
+      "step": 15340
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00029702022963367964,
+      "loss": 0.3426,
+      "step": 15350
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0002968835429196282,
+      "loss": 0.2671,
+      "step": 15360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00029674685620557684,
+      "loss": 0.3833,
+      "step": 15370
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0002966101694915254,
+      "loss": 0.3196,
+      "step": 15380
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00029647348277747404,
+      "loss": 0.3031,
+      "step": 15390
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0002963367960634226,
+      "loss": 0.3166,
+      "step": 15400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00029620010934937124,
+      "loss": 0.3602,
+      "step": 15410
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00029606342263531986,
+      "loss": 0.2904,
+      "step": 15420
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00029592673592126844,
+      "loss": 0.2737,
+      "step": 15430
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00029579004920721706,
+      "loss": 0.2722,
+      "step": 15440
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00029565336249316563,
+      "loss": 0.3298,
+      "step": 15450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00029551667577911426,
+      "loss": 0.2949,
+      "step": 15460
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00029537998906506283,
+      "loss": 0.3151,
+      "step": 15470
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0002952433023510115,
+      "loss": 0.3545,
+      "step": 15480
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00029510661563696014,
+      "loss": 0.3366,
+      "step": 15490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0002949699289229087,
+      "loss": 0.327,
+      "step": 15500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00029483324220885734,
+      "loss": 0.2797,
+      "step": 15510
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0002946965554948059,
+      "loss": 0.3526,
+      "step": 15520
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00029455986878075454,
+      "loss": 0.3261,
+      "step": 15530
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0002944231820667031,
+      "loss": 0.3667,
+      "step": 15540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00029428649535265174,
+      "loss": 0.3012,
+      "step": 15550
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00029414980863860037,
+      "loss": 0.3291,
+      "step": 15560
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00029401312192454894,
+      "loss": 0.4251,
+      "step": 15570
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00029387643521049757,
+      "loss": 0.2873,
+      "step": 15580
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00029373974849644614,
+      "loss": 0.3121,
+      "step": 15590
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00029360306178239477,
+      "loss": 0.36,
+      "step": 15600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00029346637506834334,
+      "loss": 0.2857,
+      "step": 15610
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00029332968835429197,
+      "loss": 0.3299,
+      "step": 15620
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002931930016402406,
+      "loss": 0.3096,
+      "step": 15630
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00029305631492618917,
+      "loss": 0.2952,
+      "step": 15640
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002929196282121378,
+      "loss": 0.3329,
+      "step": 15650
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00029278294149808636,
+      "loss": 0.3383,
+      "step": 15660
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.000292646254784035,
+      "loss": 0.402,
+      "step": 15670
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00029250956806998356,
+      "loss": 0.3488,
+      "step": 15680
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0002923728813559322,
+      "loss": 0.3237,
+      "step": 15690
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0002922361946418808,
+      "loss": 0.3154,
+      "step": 15700
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0002920995079278294,
+      "loss": 0.3942,
+      "step": 15710
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000291962821213778,
+      "loss": 0.3599,
+      "step": 15720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0002918261344997266,
+      "loss": 0.3683,
+      "step": 15730
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0002916894477856752,
+      "loss": 0.3667,
+      "step": 15740
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00029155276107162384,
+      "loss": 0.3192,
+      "step": 15750
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00029141607435757247,
+      "loss": 0.3926,
+      "step": 15760
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0002912793876435211,
+      "loss": 0.3994,
+      "step": 15770
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00029114270092946967,
+      "loss": 0.2871,
+      "step": 15780
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0002910060142154183,
+      "loss": 0.353,
+      "step": 15790
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00029086932750136687,
+      "loss": 0.3026,
+      "step": 15800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0002907326407873155,
+      "loss": 0.3209,
+      "step": 15810
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00029059595407326407,
+      "loss": 0.293,
+      "step": 15820
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0002904592673592127,
+      "loss": 0.307,
+      "step": 15830
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0002903225806451613,
+      "loss": 0.3246,
+      "step": 15840
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0002901858939311099,
+      "loss": 0.3179,
+      "step": 15850
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002900492072170585,
+      "loss": 0.3365,
+      "step": 15860
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002899125205030071,
+      "loss": 0.3009,
+      "step": 15870
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002897758337889557,
+      "loss": 0.2569,
+      "step": 15880
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00028963914707490435,
+      "loss": 0.2964,
+      "step": 15890
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0002895024603608529,
+      "loss": 0.3558,
+      "step": 15900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00028936577364680155,
+      "loss": 0.329,
+      "step": 15910
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0002892290869327501,
+      "loss": 0.2982,
+      "step": 15920
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00028909240021869875,
+      "loss": 0.3301,
+      "step": 15930
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0002889557135046473,
+      "loss": 0.3483,
+      "step": 15940
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00028881902679059595,
+      "loss": 0.3125,
+      "step": 15950
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0002886823400765446,
+      "loss": 0.3245,
+      "step": 15960
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00028854565336249315,
+      "loss": 0.3276,
+      "step": 15970
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0002884089666484418,
+      "loss": 0.3481,
+      "step": 15980
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00028827227993439035,
+      "loss": 0.2811,
+      "step": 15990
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00028813559322033897,
+      "loss": 0.3334,
+      "step": 16000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0002879989065062876,
+      "loss": 0.2934,
+      "step": 16010
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0002878622197922362,
+      "loss": 0.2801,
+      "step": 16020
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00028772553307818485,
+      "loss": 0.2753,
+      "step": 16030
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0002875888463641334,
+      "loss": 0.3538,
+      "step": 16040
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00028745215965008205,
+      "loss": 0.3722,
+      "step": 16050
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0002873154729360306,
+      "loss": 0.3052,
+      "step": 16060
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00028717878622197925,
+      "loss": 0.3672,
+      "step": 16070
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0002870420995079278,
+      "loss": 0.3066,
+      "step": 16080
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00028690541279387645,
+      "loss": 0.345,
+      "step": 16090
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0002867687260798251,
+      "loss": 0.3305,
+      "step": 16100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00028663203936577365,
+      "loss": 0.3695,
+      "step": 16110
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0002864953526517223,
+      "loss": 0.3257,
+      "step": 16120
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00028635866593767085,
+      "loss": 0.2981,
+      "step": 16130
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0002862219792236195,
+      "loss": 0.4048,
+      "step": 16140
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00028608529250956805,
+      "loss": 0.2927,
+      "step": 16150
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0002859486057955167,
+      "loss": 0.3764,
+      "step": 16160
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0002858119190814653,
+      "loss": 0.3123,
+      "step": 16170
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0002856752323674139,
+      "loss": 0.3544,
+      "step": 16180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0002855385456533625,
+      "loss": 0.3171,
+      "step": 16190
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0002854018589393111,
+      "loss": 0.3618,
+      "step": 16200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0002852651722252597,
+      "loss": 0.3357,
+      "step": 16210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0002851284855112083,
+      "loss": 0.3932,
+      "step": 16220
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0002849917987971569,
+      "loss": 0.3245,
+      "step": 16230
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00028485511208310553,
+      "loss": 0.3258,
+      "step": 16240
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0002847184253690541,
+      "loss": 0.3618,
+      "step": 16250
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00028458173865500273,
+      "loss": 0.3222,
+      "step": 16260
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0002844450519409513,
+      "loss": 0.3832,
+      "step": 16270
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0002843083652269,
+      "loss": 0.3899,
+      "step": 16280
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00028417167851284855,
+      "loss": 0.3239,
+      "step": 16290
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0002840349917987972,
+      "loss": 0.3808,
+      "step": 16300
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0002838983050847458,
+      "loss": 0.3043,
+      "step": 16310
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0002837616183706944,
+      "loss": 0.3598,
+      "step": 16320
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.000283624931656643,
+      "loss": 0.3658,
+      "step": 16330
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002834882449425916,
+      "loss": 0.2951,
+      "step": 16340
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002833515582285402,
+      "loss": 0.3938,
+      "step": 16350
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002832148715144888,
+      "loss": 0.3607,
+      "step": 16360
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002830781848004374,
+      "loss": 0.3,
+      "step": 16370
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00028294149808638603,
+      "loss": 0.3481,
+      "step": 16380
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0002828048113723346,
+      "loss": 0.3039,
+      "step": 16390
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00028266812465828323,
+      "loss": 0.3939,
+      "step": 16400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0002825314379442318,
+      "loss": 0.3222,
+      "step": 16410
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00028239475123018043,
+      "loss": 0.306,
+      "step": 16420
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00028225806451612906,
+      "loss": 0.3612,
+      "step": 16430
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00028212137780207763,
+      "loss": 0.3203,
+      "step": 16440
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00028198469108802626,
+      "loss": 0.3704,
+      "step": 16450
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00028184800437397483,
+      "loss": 0.3254,
+      "step": 16460
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00028171131765992346,
+      "loss": 0.3067,
+      "step": 16470
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00028157463094587203,
+      "loss": 0.3095,
+      "step": 16480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00028143794423182066,
+      "loss": 0.3348,
+      "step": 16490
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0002813012575177693,
+      "loss": 0.3033,
+      "step": 16500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00028116457080371786,
+      "loss": 0.3588,
+      "step": 16510
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0002810278840896665,
+      "loss": 0.3074,
+      "step": 16520
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00028089119737561506,
+      "loss": 0.3621,
+      "step": 16530
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00028075451066156374,
+      "loss": 0.2925,
+      "step": 16540
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0002806178239475123,
+      "loss": 0.3629,
+      "step": 16550
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00028048113723346094,
+      "loss": 0.3785,
+      "step": 16560
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00028034445051940956,
+      "loss": 0.3126,
+      "step": 16570
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00028020776380535814,
+      "loss": 0.2964,
+      "step": 16580
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00028007107709130676,
+      "loss": 0.2821,
+      "step": 16590
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00027993439037725534,
+      "loss": 0.3843,
+      "step": 16600
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00027979770366320396,
+      "loss": 0.3506,
+      "step": 16610
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00027966101694915254,
+      "loss": 0.3204,
+      "step": 16620
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00027952433023510116,
+      "loss": 0.3423,
+      "step": 16630
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0002793876435210498,
+      "loss": 0.3892,
+      "step": 16640
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00027925095680699836,
+      "loss": 0.3787,
+      "step": 16650
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000279114270092947,
+      "loss": 0.3625,
+      "step": 16660
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00027897758337889556,
+      "loss": 0.3551,
+      "step": 16670
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0002788408966648442,
+      "loss": 0.2586,
+      "step": 16680
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00027870420995079276,
+      "loss": 0.4021,
+      "step": 16690
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0002785675232367414,
+      "loss": 0.2957,
+      "step": 16700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00027843083652269,
+      "loss": 0.3622,
+      "step": 16710
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0002782941498086386,
+      "loss": 0.3466,
+      "step": 16720
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0002781574630945872,
+      "loss": 0.3712,
+      "step": 16730
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0002780207763805358,
+      "loss": 0.341,
+      "step": 16740
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0002778840896664844,
+      "loss": 0.367,
+      "step": 16750
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.000277747402952433,
+      "loss": 0.3467,
+      "step": 16760
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0002776107162383816,
+      "loss": 0.3669,
+      "step": 16770
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00027747402952433024,
+      "loss": 0.3742,
+      "step": 16780
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0002773373428102788,
+      "loss": 0.323,
+      "step": 16790
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00027720065609622744,
+      "loss": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00027706396938217607,
+      "loss": 0.3925,
+      "step": 16810
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0002769272826681247,
+      "loss": 0.352,
+      "step": 16820
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00027679059595407327,
+      "loss": 0.4478,
+      "step": 16830
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0002766539092400219,
+      "loss": 0.3469,
+      "step": 16840
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0002765172225259705,
+      "loss": 0.3621,
+      "step": 16850
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0002763805358119191,
+      "loss": 0.3414,
+      "step": 16860
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0002762438490978677,
+      "loss": 0.307,
+      "step": 16870
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0002761071623838163,
+      "loss": 0.3295,
+      "step": 16880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002759704756697649,
+      "loss": 0.371,
+      "step": 16890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002758337889557135,
+      "loss": 0.3129,
+      "step": 16900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002756971022416621,
+      "loss": 0.3699,
+      "step": 16910
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00027556041552761074,
+      "loss": 0.3214,
+      "step": 16920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0002754237288135593,
+      "loss": 0.3817,
+      "step": 16930
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00027528704209950794,
+      "loss": 0.3621,
+      "step": 16940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0002751503553854565,
+      "loss": 0.3595,
+      "step": 16950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00027501366867140514,
+      "loss": 0.3272,
+      "step": 16960
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00027487698195735377,
+      "loss": 0.3431,
+      "step": 16970
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00027474029524330234,
+      "loss": 0.3541,
+      "step": 16980
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00027460360852925097,
+      "loss": 0.3316,
+      "step": 16990
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00027446692181519954,
+      "loss": 0.3447,
+      "step": 17000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00027433023510114817,
+      "loss": 0.3189,
+      "step": 17010
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00027419354838709674,
+      "loss": 0.3114,
+      "step": 17020
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00027405686167304537,
+      "loss": 0.4111,
+      "step": 17030
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.000273920174958994,
+      "loss": 0.3706,
+      "step": 17040
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00027378348824494257,
+      "loss": 0.3392,
+      "step": 17050
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0002736468015308912,
+      "loss": 0.4149,
+      "step": 17060
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0002735101148168398,
+      "loss": 0.3268,
+      "step": 17070
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00027337342810278845,
+      "loss": 0.3213,
+      "step": 17080
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000273236741388737,
+      "loss": 0.3219,
+      "step": 17090
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00027310005467468565,
+      "loss": 0.3243,
+      "step": 17100
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0002729633679606343,
+      "loss": 0.3623,
+      "step": 17110
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00027282668124658285,
+      "loss": 0.3752,
+      "step": 17120
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0002726899945325315,
+      "loss": 0.3272,
+      "step": 17130
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00027255330781848005,
+      "loss": 0.3253,
+      "step": 17140
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0002724166211044287,
+      "loss": 0.4001,
+      "step": 17150
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00027227993439037725,
+      "loss": 0.3323,
+      "step": 17160
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0002721432476763259,
+      "loss": 0.3341,
+      "step": 17170
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0002720065609622745,
+      "loss": 0.3645,
+      "step": 17180
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00027186987424822307,
+      "loss": 0.4015,
+      "step": 17190
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0002717331875341717,
+      "loss": 0.4092,
+      "step": 17200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00027159650082012027,
+      "loss": 0.3704,
+      "step": 17210
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0002714598141060689,
+      "loss": 0.3094,
+      "step": 17220
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00027132312739201747,
+      "loss": 0.3628,
+      "step": 17230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0002711864406779661,
+      "loss": 0.3282,
+      "step": 17240
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0002710497539639147,
+      "loss": 0.3434,
+      "step": 17250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0002709130672498633,
+      "loss": 0.3977,
+      "step": 17260
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0002707763805358119,
+      "loss": 0.3824,
+      "step": 17270
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0002706396938217605,
+      "loss": 0.349,
+      "step": 17280
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0002705030071077091,
+      "loss": 0.3792,
+      "step": 17290
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002703663203936577,
+      "loss": 0.2978,
+      "step": 17300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002702296336796063,
+      "loss": 0.3523,
+      "step": 17310
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00027009294696555495,
+      "loss": 0.3895,
+      "step": 17320
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002699562602515035,
+      "loss": 0.3632,
+      "step": 17330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0002698195735374522,
+      "loss": 0.35,
+      "step": 17340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0002696828868234008,
+      "loss": 0.2886,
+      "step": 17350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0002695462001093494,
+      "loss": 0.3288,
+      "step": 17360
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.000269409513395298,
+      "loss": 0.3664,
+      "step": 17370
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0002692728266812466,
+      "loss": 0.3564,
+      "step": 17380
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00026913613996719523,
+      "loss": 0.3481,
+      "step": 17390
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0002689994532531438,
+      "loss": 0.3208,
+      "step": 17400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00026886276653909243,
+      "loss": 0.3242,
+      "step": 17410
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.000268726079825041,
+      "loss": 0.3724,
+      "step": 17420
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00026858939311098963,
+      "loss": 0.4079,
+      "step": 17430
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0002684527063969382,
+      "loss": 0.3121,
+      "step": 17440
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00026831601968288683,
+      "loss": 0.352,
+      "step": 17450
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00026817933296883545,
+      "loss": 0.3202,
+      "step": 17460
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00026804264625478403,
+      "loss": 0.3539,
+      "step": 17470
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00026790595954073265,
+      "loss": 0.3891,
+      "step": 17480
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0002677692728266812,
+      "loss": 0.2785,
+      "step": 17490
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00026763258611262985,
+      "loss": 0.3556,
+      "step": 17500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0002674958993985785,
+      "loss": 0.3466,
+      "step": 17510
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00026735921268452705,
+      "loss": 0.341,
+      "step": 17520
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0002672225259704757,
+      "loss": 0.3322,
+      "step": 17530
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00026708583925642425,
+      "loss": 0.3404,
+      "step": 17540
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0002669491525423729,
+      "loss": 0.3804,
+      "step": 17550
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00026681246582832145,
+      "loss": 0.3155,
+      "step": 17560
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002666757791142701,
+      "loss": 0.3624,
+      "step": 17570
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002665390924002187,
+      "loss": 0.3543,
+      "step": 17580
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002664024056861673,
+      "loss": 0.3657,
+      "step": 17590
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00026626571897211596,
+      "loss": 0.388,
+      "step": 17600
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00026612903225806453,
+      "loss": 0.3234,
+      "step": 17610
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00026599234554401316,
+      "loss": 0.3799,
+      "step": 17620
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00026585565882996173,
+      "loss": 0.341,
+      "step": 17630
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00026571897211591036,
+      "loss": 0.3522,
+      "step": 17640
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000265582285401859,
+      "loss": 0.3565,
+      "step": 17650
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00026544559868780756,
+      "loss": 0.3656,
+      "step": 17660
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0002653089119737562,
+      "loss": 0.3502,
+      "step": 17670
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00026517222525970476,
+      "loss": 0.3505,
+      "step": 17680
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0002650355385456534,
+      "loss": 0.3291,
+      "step": 17690
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00026489885183160196,
+      "loss": 0.3411,
+      "step": 17700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0002647621651175506,
+      "loss": 0.3467,
+      "step": 17710
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0002646254784034992,
+      "loss": 0.3533,
+      "step": 17720
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0002644887916894478,
+      "loss": 0.3695,
+      "step": 17730
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0002643521049753964,
+      "loss": 0.3429,
+      "step": 17740
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000264215418261345,
+      "loss": 0.3253,
+      "step": 17750
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0002640787315472936,
+      "loss": 0.3645,
+      "step": 17760
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0002639420448332422,
+      "loss": 0.3523,
+      "step": 17770
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0002638053581191908,
+      "loss": 0.3715,
+      "step": 17780
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00026366867140513944,
+      "loss": 0.3308,
+      "step": 17790
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.000263531984691088,
+      "loss": 0.3753,
+      "step": 17800
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00026339529797703664,
+      "loss": 0.3794,
+      "step": 17810
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0002632586112629852,
+      "loss": 0.3675,
+      "step": 17820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00026312192454893383,
+      "loss": 0.2962,
+      "step": 17830
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0002629852378348824,
+      "loss": 0.4128,
+      "step": 17840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00026284855112083103,
+      "loss": 0.3403,
+      "step": 17850
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0002627118644067797,
+      "loss": 0.3924,
+      "step": 17860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0002625751776927283,
+      "loss": 0.3065,
+      "step": 17870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0002624384909786769,
+      "loss": 0.3555,
+      "step": 17880
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0002623018042646255,
+      "loss": 0.4393,
+      "step": 17890
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0002621651175505741,
+      "loss": 0.3475,
+      "step": 17900
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0002620284308365227,
+      "loss": 0.2872,
+      "step": 17910
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0002618917441224713,
+      "loss": 0.3365,
+      "step": 17920
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00026175505740841994,
+      "loss": 0.4278,
+      "step": 17930
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0002616183706943685,
+      "loss": 0.3403,
+      "step": 17940
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00026148168398031714,
+      "loss": 0.3583,
+      "step": 17950
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0002613449972662657,
+      "loss": 0.3577,
+      "step": 17960
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00026120831055221434,
+      "loss": 0.3555,
+      "step": 17970
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0002610716238381629,
+      "loss": 0.3513,
+      "step": 17980
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00026093493712411154,
+      "loss": 0.3945,
+      "step": 17990
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00026079825041006017,
+      "loss": 0.3353,
+      "step": 18000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00026066156369600874,
+      "loss": 0.3202,
+      "step": 18010
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00026052487698195737,
+      "loss": 0.2841,
+      "step": 18020
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00026038819026790594,
+      "loss": 0.3455,
+      "step": 18030
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00026025150355385456,
+      "loss": 0.3361,
+      "step": 18040
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0002601148168398032,
+      "loss": 0.3698,
+      "step": 18050
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00025997813012575176,
+      "loss": 0.3917,
+      "step": 18060
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0002598414434117004,
+      "loss": 0.3362,
+      "step": 18070
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00025970475669764896,
+      "loss": 0.2967,
+      "step": 18080
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0002595680699835976,
+      "loss": 0.3644,
+      "step": 18090
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00025943138326954616,
+      "loss": 0.3746,
+      "step": 18100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0002592946965554948,
+      "loss": 0.3905,
+      "step": 18110
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0002591580098414434,
+      "loss": 0.3271,
+      "step": 18120
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00025902132312739204,
+      "loss": 0.4043,
+      "step": 18130
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00025888463641334067,
+      "loss": 0.4111,
+      "step": 18140
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00025874794969928924,
+      "loss": 0.3594,
+      "step": 18150
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00025861126298523787,
+      "loss": 0.3639,
+      "step": 18160
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00025847457627118644,
+      "loss": 0.3512,
+      "step": 18170
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00025833788955713507,
+      "loss": 0.4299,
+      "step": 18180
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0002582012028430837,
+      "loss": 0.3713,
+      "step": 18190
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00025806451612903227,
+      "loss": 0.407,
+      "step": 18200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0002579278294149809,
+      "loss": 0.3616,
+      "step": 18210
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00025779114270092947,
+      "loss": 0.3485,
+      "step": 18220
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0002576544559868781,
+      "loss": 0.3361,
+      "step": 18230
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00025751776927282667,
+      "loss": 0.3398,
+      "step": 18240
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0002573810825587753,
+      "loss": 0.3627,
+      "step": 18250
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0002572443958447239,
+      "loss": 0.3778,
+      "step": 18260
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002571077091306725,
+      "loss": 0.3559,
+      "step": 18270
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002569710224166211,
+      "loss": 0.3855,
+      "step": 18280
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002568343357025697,
+      "loss": 0.3777,
+      "step": 18290
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002566976489885183,
+      "loss": 0.3743,
+      "step": 18300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002565609622744669,
+      "loss": 0.3797,
+      "step": 18310
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002564242755604155,
+      "loss": 0.4019,
+      "step": 18320
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00025628758884636415,
+      "loss": 0.4332,
+      "step": 18330
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0002561509021323127,
+      "loss": 0.3682,
+      "step": 18340
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00025601421541826135,
+      "loss": 0.3397,
+      "step": 18350
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0002558775287042099,
+      "loss": 0.3333,
+      "step": 18360
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00025574084199015855,
+      "loss": 0.3429,
+      "step": 18370
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0002556041552761071,
+      "loss": 0.3068,
+      "step": 18380
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00025546746856205574,
+      "loss": 0.3422,
+      "step": 18390
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0002553307818480044,
+      "loss": 0.3766,
+      "step": 18400
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.000255194095133953,
+      "loss": 0.338,
+      "step": 18410
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002550574084199016,
+      "loss": 0.4689,
+      "step": 18420
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002549207217058502,
+      "loss": 0.3729,
+      "step": 18430
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002547840349917988,
+      "loss": 0.3393,
+      "step": 18440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0002546473482777474,
+      "loss": 0.3394,
+      "step": 18450
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.000254510661563696,
+      "loss": 0.3882,
+      "step": 18460
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00025437397484964465,
+      "loss": 0.3873,
+      "step": 18470
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0002542372881355932,
+      "loss": 0.3553,
+      "step": 18480
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00025410060142154185,
+      "loss": 0.3458,
+      "step": 18490
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0002539639147074904,
+      "loss": 0.3935,
+      "step": 18500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00025382722799343905,
+      "loss": 0.3937,
+      "step": 18510
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0002536905412793876,
+      "loss": 0.373,
+      "step": 18520
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00025355385456533625,
+      "loss": 0.3503,
+      "step": 18530
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0002534171678512849,
+      "loss": 0.3513,
+      "step": 18540
+    },
+    {
+      "epoch": 5.0,
+      "eval_Bleu_1": 0.004668591436820157,
+      "eval_Bleu_2": 1.9669498094375504e-11,
+      "eval_Bleu_3": 3.5057923977893674e-14,
+      "eval_Bleu_4": 1.5738342023347648e-15,
+      "eval_ROUGE_L": 0.007040352180116771,
+      "eval_cer": 0.6059977605764082,
+      "eval_em": 0.4442070275403609,
+      "eval_f1": 0.6063440222377056,
+      "eval_loss": 0.8440435528755188,
+      "eval_runtime": 316.8318,
+      "eval_samples_per_second": 13.294,
+      "eval_steps_per_second": 2.661,
+      "eval_wer": 0.7139830508474576,
+      "step": 18540
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00025328048113723345,
+      "loss": 0.2585,
+      "step": 18550
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002531437944231821,
+      "loss": 0.2126,
+      "step": 18560
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00025300710770913065,
+      "loss": 0.2334,
+      "step": 18570
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002528704209950793,
+      "loss": 0.2135,
+      "step": 18580
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002527337342810279,
+      "loss": 0.2439,
+      "step": 18590
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002525970475669765,
+      "loss": 0.2201,
+      "step": 18600
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002524603608529251,
+      "loss": 0.251,
+      "step": 18610
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002523236741388737,
+      "loss": 0.2643,
+      "step": 18620
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002521869874248223,
+      "loss": 0.2282,
+      "step": 18630
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002520503007107709,
+      "loss": 0.1794,
+      "step": 18640
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002519136139967195,
+      "loss": 0.2169,
+      "step": 18650
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002517769272826682,
+      "loss": 0.2485,
+      "step": 18660
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00025164024056861675,
+      "loss": 0.2049,
+      "step": 18670
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002515035538545654,
+      "loss": 0.2526,
+      "step": 18680
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00025136686714051395,
+      "loss": 0.2403,
+      "step": 18690
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002512301804264626,
+      "loss": 0.248,
+      "step": 18700
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00025109349371241115,
+      "loss": 0.2524,
+      "step": 18710
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002509568069983598,
+      "loss": 0.1957,
+      "step": 18720
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002508201202843084,
+      "loss": 0.2659,
+      "step": 18730
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.000250683433570257,
+      "loss": 0.2164,
+      "step": 18740
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002505467468562056,
+      "loss": 0.1919,
+      "step": 18750
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002504100601421542,
+      "loss": 0.2481,
+      "step": 18760
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002502733734281028,
+      "loss": 0.2146,
+      "step": 18770
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002501366867140514,
+      "loss": 0.2403,
+      "step": 18780
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00025,
+      "loss": 0.1862,
+      "step": 18790
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00024986331328594863,
+      "loss": 0.2042,
+      "step": 18800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002497266265718972,
+      "loss": 0.2755,
+      "step": 18810
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00024958993985784583,
+      "loss": 0.2287,
+      "step": 18820
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00024945325314379446,
+      "loss": 0.2632,
+      "step": 18830
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00024931656642974303,
+      "loss": 0.2497,
+      "step": 18840
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00024917987971569166,
+      "loss": 0.2031,
+      "step": 18850
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00024904319300164023,
+      "loss": 0.2506,
+      "step": 18860
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00024890650628758886,
+      "loss": 0.2803,
+      "step": 18870
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00024876981957353743,
+      "loss": 0.2503,
+      "step": 18880
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00024863313285948606,
+      "loss": 0.2207,
+      "step": 18890
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002484964461454347,
+      "loss": 0.2457,
+      "step": 18900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00024835975943138326,
+      "loss": 0.2106,
+      "step": 18910
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002482230727173319,
+      "loss": 0.2653,
+      "step": 18920
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002480863860032805,
+      "loss": 0.2353,
+      "step": 18930
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002479496992892291,
+      "loss": 0.2319,
+      "step": 18940
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002478130125751777,
+      "loss": 0.3066,
+      "step": 18950
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00024767632586112634,
+      "loss": 0.2945,
+      "step": 18960
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002475396391470749,
+      "loss": 0.2417,
+      "step": 18970
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00024740295243302354,
+      "loss": 0.2767,
+      "step": 18980
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002472662657189721,
+      "loss": 0.2058,
+      "step": 18990
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00024712957900492074,
+      "loss": 0.1913,
+      "step": 19000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002469928922908693,
+      "loss": 0.2059,
+      "step": 19010
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00024685620557681793,
+      "loss": 0.2184,
+      "step": 19020
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00024671951886276656,
+      "loss": 0.2049,
+      "step": 19030
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00024658283214871513,
+      "loss": 0.2682,
+      "step": 19040
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00024644614543466376,
+      "loss": 0.2052,
+      "step": 19050
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002463094587206124,
+      "loss": 0.2319,
+      "step": 19060
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00024617277200656096,
+      "loss": 0.2834,
+      "step": 19070
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002460360852925096,
+      "loss": 0.2491,
+      "step": 19080
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00024589939857845816,
+      "loss": 0.2035,
+      "step": 19090
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002457627118644068,
+      "loss": 0.2299,
+      "step": 19100
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002456260251503554,
+      "loss": 0.2183,
+      "step": 19110
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.000245489338436304,
+      "loss": 0.2411,
+      "step": 19120
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002453526517222526,
+      "loss": 0.2702,
+      "step": 19130
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002452159650082012,
+      "loss": 0.2923,
+      "step": 19140
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002450792782941498,
+      "loss": 0.2331,
+      "step": 19150
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002449425915800984,
+      "loss": 0.2241,
+      "step": 19160
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.000244805904866047,
+      "loss": 0.2128,
+      "step": 19170
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00024466921815199564,
+      "loss": 0.2535,
+      "step": 19180
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00024453253143794427,
+      "loss": 0.2369,
+      "step": 19190
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00024439584472389284,
+      "loss": 0.2781,
+      "step": 19200
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00024425915800984146,
+      "loss": 0.2399,
+      "step": 19210
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00024412247129579004,
+      "loss": 0.2458,
+      "step": 19220
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00024398578458173866,
+      "loss": 0.2499,
+      "step": 19230
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00024384909786768726,
+      "loss": 0.2089,
+      "step": 19240
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00024371241115363586,
+      "loss": 0.2193,
+      "step": 19250
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002435757244395845,
+      "loss": 0.1923,
+      "step": 19260
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002434390377255331,
+      "loss": 0.2419,
+      "step": 19270
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002433023510114817,
+      "loss": 0.2534,
+      "step": 19280
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002431656642974303,
+      "loss": 0.2318,
+      "step": 19290
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00024302897758337892,
+      "loss": 0.2507,
+      "step": 19300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00024289229086932752,
+      "loss": 0.2189,
+      "step": 19310
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00024275560415527612,
+      "loss": 0.1964,
+      "step": 19320
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00024261891744122472,
+      "loss": 0.2277,
+      "step": 19330
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00024248223072717332,
+      "loss": 0.23,
+      "step": 19340
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00024234554401312192,
+      "loss": 0.2539,
+      "step": 19350
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00024220885729907052,
+      "loss": 0.1781,
+      "step": 19360
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00024207217058501914,
+      "loss": 0.2933,
+      "step": 19370
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00024193548387096774,
+      "loss": 0.242,
+      "step": 19380
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00024179879715691637,
+      "loss": 0.2764,
+      "step": 19390
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00024166211044286497,
+      "loss": 0.2599,
+      "step": 19400
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00024152542372881357,
+      "loss": 0.2297,
+      "step": 19410
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00024138873701476217,
+      "loss": 0.2645,
+      "step": 19420
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00024125205030071077,
+      "loss": 0.2075,
+      "step": 19430
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002411153635866594,
+      "loss": 0.2457,
+      "step": 19440
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.000240978676872608,
+      "loss": 0.2204,
+      "step": 19450
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002408419901585566,
+      "loss": 0.224,
+      "step": 19460
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002407053034445052,
+      "loss": 0.243,
+      "step": 19470
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002405686167304538,
+      "loss": 0.3034,
+      "step": 19480
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002404319300164024,
+      "loss": 0.2841,
+      "step": 19490
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00024029524330235102,
+      "loss": 0.2409,
+      "step": 19500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00024015855658829962,
+      "loss": 0.2896,
+      "step": 19510
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00024002186987424825,
+      "loss": 0.2294,
+      "step": 19520
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00023988518316019685,
+      "loss": 0.2459,
+      "step": 19530
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00023974849644614545,
+      "loss": 0.2146,
+      "step": 19540
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00023961180973209405,
+      "loss": 0.258,
+      "step": 19550
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00023947512301804265,
+      "loss": 0.2563,
+      "step": 19560
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00023933843630399127,
+      "loss": 0.2803,
+      "step": 19570
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00023920174958993987,
+      "loss": 0.2468,
+      "step": 19580
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00023906506287588847,
+      "loss": 0.2841,
+      "step": 19590
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00023892837616183707,
+      "loss": 0.2426,
+      "step": 19600
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00023879168944778567,
+      "loss": 0.2075,
+      "step": 19610
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00023865500273373427,
+      "loss": 0.2569,
+      "step": 19620
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00023851831601968287,
+      "loss": 0.2502,
+      "step": 19630
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002383816293056315,
+      "loss": 0.2437,
+      "step": 19640
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002382449425915801,
+      "loss": 0.2025,
+      "step": 19650
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00023810825587752872,
+      "loss": 0.2289,
+      "step": 19660
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00023797156916347732,
+      "loss": 0.2356,
+      "step": 19670
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00023783488244942592,
+      "loss": 0.2644,
+      "step": 19680
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00023769819573537452,
+      "loss": 0.2601,
+      "step": 19690
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00023756150902132312,
+      "loss": 0.2319,
+      "step": 19700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00023742482230727175,
+      "loss": 0.2451,
+      "step": 19710
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00023728813559322035,
+      "loss": 0.2616,
+      "step": 19720
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00023715144887916895,
+      "loss": 0.224,
+      "step": 19730
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00023701476216511755,
+      "loss": 0.2641,
+      "step": 19740
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00023687807545106615,
+      "loss": 0.3065,
+      "step": 19750
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00023674138873701475,
+      "loss": 0.2224,
+      "step": 19760
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00023660470202296338,
+      "loss": 0.2872,
+      "step": 19770
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00023646801530891197,
+      "loss": 0.2594,
+      "step": 19780
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002363313285948606,
+      "loss": 0.2579,
+      "step": 19790
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002361946418808092,
+      "loss": 0.166,
+      "step": 19800
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002360579551667578,
+      "loss": 0.2095,
+      "step": 19810
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002359212684527064,
+      "loss": 0.2419,
+      "step": 19820
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.000235784581738655,
+      "loss": 0.2441,
+      "step": 19830
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00023564789502460363,
+      "loss": 0.2819,
+      "step": 19840
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00023551120831055223,
+      "loss": 0.2555,
+      "step": 19850
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00023537452159650083,
+      "loss": 0.2842,
+      "step": 19860
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00023523783488244943,
+      "loss": 0.2751,
+      "step": 19870
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00023510114816839803,
+      "loss": 0.2325,
+      "step": 19880
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00023496446145434663,
+      "loss": 0.2453,
+      "step": 19890
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00023482777474029523,
+      "loss": 0.2733,
+      "step": 19900
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00023469108802624385,
+      "loss": 0.2341,
+      "step": 19910
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00023455440131219248,
+      "loss": 0.2233,
+      "step": 19920
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00023441771459814108,
+      "loss": 0.2639,
+      "step": 19930
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00023428102788408968,
+      "loss": 0.2623,
+      "step": 19940
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00023414434117003828,
+      "loss": 0.2869,
+      "step": 19950
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00023400765445598688,
+      "loss": 0.2442,
+      "step": 19960
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00023387096774193548,
+      "loss": 0.2849,
+      "step": 19970
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002337342810278841,
+      "loss": 0.2864,
+      "step": 19980
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002335975943138327,
+      "loss": 0.2004,
+      "step": 19990
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002334609075997813,
+      "loss": 0.2449,
+      "step": 20000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002333242208857299,
+      "loss": 0.2024,
+      "step": 20010
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002331875341716785,
+      "loss": 0.2862,
+      "step": 20020
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002330508474576271,
+      "loss": 0.233,
+      "step": 20030
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00023291416074357573,
+      "loss": 0.2521,
+      "step": 20040
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00023277747402952436,
+      "loss": 0.3056,
+      "step": 20050
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00023264078731547296,
+      "loss": 0.2054,
+      "step": 20060
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00023250410060142156,
+      "loss": 0.2631,
+      "step": 20070
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00023236741388737016,
+      "loss": 0.2562,
+      "step": 20080
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00023223072717331876,
+      "loss": 0.254,
+      "step": 20090
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00023209404045926736,
+      "loss": 0.2134,
+      "step": 20100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00023195735374521598,
+      "loss": 0.2256,
+      "step": 20110
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00023182066703116458,
+      "loss": 0.2482,
+      "step": 20120
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00023168398031711318,
+      "loss": 0.2594,
+      "step": 20130
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00023154729360306178,
+      "loss": 0.2055,
+      "step": 20140
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00023141060688901038,
+      "loss": 0.2302,
+      "step": 20150
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00023127392017495898,
+      "loss": 0.2522,
+      "step": 20160
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00023113723346090758,
+      "loss": 0.274,
+      "step": 20170
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002310005467468562,
+      "loss": 0.1899,
+      "step": 20180
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00023086386003280483,
+      "loss": 0.233,
+      "step": 20190
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00023072717331875343,
+      "loss": 0.209,
+      "step": 20200
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00023059048660470203,
+      "loss": 0.1983,
+      "step": 20210
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00023045379989065063,
+      "loss": 0.2781,
+      "step": 20220
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00023031711317659923,
+      "loss": 0.2635,
+      "step": 20230
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00023018042646254783,
+      "loss": 0.2355,
+      "step": 20240
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00023004373974849646,
+      "loss": 0.278,
+      "step": 20250
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00022990705303444506,
+      "loss": 0.2494,
+      "step": 20260
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00022977036632039366,
+      "loss": 0.2592,
+      "step": 20270
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00022963367960634226,
+      "loss": 0.2915,
+      "step": 20280
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00022949699289229086,
+      "loss": 0.2555,
+      "step": 20290
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00022936030617823946,
+      "loss": 0.2727,
+      "step": 20300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00022922361946418809,
+      "loss": 0.1963,
+      "step": 20310
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002290869327501367,
+      "loss": 0.2555,
+      "step": 20320
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002289502460360853,
+      "loss": 0.2319,
+      "step": 20330
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002288135593220339,
+      "loss": 0.2332,
+      "step": 20340
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002286768726079825,
+      "loss": 0.2693,
+      "step": 20350
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002285401858939311,
+      "loss": 0.2192,
+      "step": 20360
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002284034991798797,
+      "loss": 0.2451,
+      "step": 20370
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00022826681246582834,
+      "loss": 0.2774,
+      "step": 20380
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00022813012575177694,
+      "loss": 0.2745,
+      "step": 20390
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00022799343903772554,
+      "loss": 0.2895,
+      "step": 20400
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00022785675232367414,
+      "loss": 0.2162,
+      "step": 20410
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00022772006560962274,
+      "loss": 0.2876,
+      "step": 20420
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00022758337889557134,
+      "loss": 0.2624,
+      "step": 20430
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00022744669218151994,
+      "loss": 0.2622,
+      "step": 20440
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002273100054674686,
+      "loss": 0.3169,
+      "step": 20450
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002271733187534172,
+      "loss": 0.2834,
+      "step": 20460
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002270366320393658,
+      "loss": 0.2517,
+      "step": 20470
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002268999453253144,
+      "loss": 0.261,
+      "step": 20480
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.000226763258611263,
+      "loss": 0.2017,
+      "step": 20490
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002266265718972116,
+      "loss": 0.1934,
+      "step": 20500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002264898851831602,
+      "loss": 0.2427,
+      "step": 20510
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00022635319846910882,
+      "loss": 0.2552,
+      "step": 20520
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00022621651175505742,
+      "loss": 0.2485,
+      "step": 20530
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00022607982504100602,
+      "loss": 0.2639,
+      "step": 20540
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00022594313832695462,
+      "loss": 0.2858,
+      "step": 20550
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00022580645161290321,
+      "loss": 0.2612,
+      "step": 20560
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00022566976489885181,
+      "loss": 0.2166,
+      "step": 20570
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00022553307818480047,
+      "loss": 0.29,
+      "step": 20580
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00022539639147074907,
+      "loss": 0.2229,
+      "step": 20590
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00022525970475669767,
+      "loss": 0.3133,
+      "step": 20600
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00022512301804264627,
+      "loss": 0.2471,
+      "step": 20610
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00022498633132859487,
+      "loss": 0.2558,
+      "step": 20620
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00022484964461454347,
+      "loss": 0.2251,
+      "step": 20630
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00022471295790049207,
+      "loss": 0.2607,
+      "step": 20640
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002245762711864407,
+      "loss": 0.2445,
+      "step": 20650
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002244395844723893,
+      "loss": 0.292,
+      "step": 20660
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002243028977583379,
+      "loss": 0.3075,
+      "step": 20670
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002241662110442865,
+      "loss": 0.2727,
+      "step": 20680
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002240295243302351,
+      "loss": 0.2674,
+      "step": 20690
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002238928376161837,
+      "loss": 0.2439,
+      "step": 20700
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002237561509021323,
+      "loss": 0.2661,
+      "step": 20710
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00022361946418808095,
+      "loss": 0.2897,
+      "step": 20720
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00022348277747402955,
+      "loss": 0.3174,
+      "step": 20730
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00022334609075997815,
+      "loss": 0.2081,
+      "step": 20740
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00022320940404592675,
+      "loss": 0.2801,
+      "step": 20750
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00022307271733187534,
+      "loss": 0.2179,
+      "step": 20760
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00022293603061782394,
+      "loss": 0.2374,
+      "step": 20770
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00022279934390377254,
+      "loss": 0.2611,
+      "step": 20780
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00022266265718972117,
+      "loss": 0.2561,
+      "step": 20790
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00022252597047566977,
+      "loss": 0.25,
+      "step": 20800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00022238928376161837,
+      "loss": 0.2493,
+      "step": 20810
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00022225259704756697,
+      "loss": 0.2638,
+      "step": 20820
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00022211591033351557,
+      "loss": 0.2358,
+      "step": 20830
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00022197922361946417,
+      "loss": 0.2719,
+      "step": 20840
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00022184253690541282,
+      "loss": 0.2509,
+      "step": 20850
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00022170585019136142,
+      "loss": 0.2223,
+      "step": 20860
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00022156916347731002,
+      "loss": 0.2705,
+      "step": 20870
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00022143247676325862,
+      "loss": 0.225,
+      "step": 20880
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00022129579004920722,
+      "loss": 0.2596,
+      "step": 20890
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00022115910333515582,
+      "loss": 0.2967,
+      "step": 20900
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00022102241662110442,
+      "loss": 0.2547,
+      "step": 20910
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00022088572990705305,
+      "loss": 0.2772,
+      "step": 20920
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00022074904319300165,
+      "loss": 0.2781,
+      "step": 20930
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00022061235647895025,
+      "loss": 0.3175,
+      "step": 20940
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00022047566976489885,
+      "loss": 0.2411,
+      "step": 20950
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00022033898305084745,
+      "loss": 0.2488,
+      "step": 20960
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00022020229633679605,
+      "loss": 0.2484,
+      "step": 20970
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00022006560962274467,
+      "loss": 0.2289,
+      "step": 20980
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002199289229086933,
+      "loss": 0.3313,
+      "step": 20990
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002197922361946419,
+      "loss": 0.2653,
+      "step": 21000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002196555494805905,
+      "loss": 0.267,
+      "step": 21010
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002195188627665391,
+      "loss": 0.2455,
+      "step": 21020
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002193821760524877,
+      "loss": 0.2368,
+      "step": 21030
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002192454893384363,
+      "loss": 0.2507,
+      "step": 21040
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002191088026243849,
+      "loss": 0.2566,
+      "step": 21050
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00021897211591033353,
+      "loss": 0.2754,
+      "step": 21060
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00021883542919628213,
+      "loss": 0.2946,
+      "step": 21070
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00021869874248223073,
+      "loss": 0.2566,
+      "step": 21080
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00021856205576817933,
+      "loss": 0.2607,
+      "step": 21090
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00021842536905412793,
+      "loss": 0.2368,
+      "step": 21100
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00021828868234007653,
+      "loss": 0.3135,
+      "step": 21110
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00021815199562602518,
+      "loss": 0.2258,
+      "step": 21120
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00021801530891197378,
+      "loss": 0.2478,
+      "step": 21130
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00021787862219792238,
+      "loss": 0.2761,
+      "step": 21140
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00021774193548387098,
+      "loss": 0.2443,
+      "step": 21150
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00021760524876981958,
+      "loss": 0.2526,
+      "step": 21160
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00021746856205576818,
+      "loss": 0.2317,
+      "step": 21170
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00021733187534171678,
+      "loss": 0.2152,
+      "step": 21180
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002171951886276654,
+      "loss": 0.2549,
+      "step": 21190
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.000217058501913614,
+      "loss": 0.252,
+      "step": 21200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002169218151995626,
+      "loss": 0.2542,
+      "step": 21210
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002167851284855112,
+      "loss": 0.2367,
+      "step": 21220
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002166484417714598,
+      "loss": 0.2438,
+      "step": 21230
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002165117550574084,
+      "loss": 0.2598,
+      "step": 21240
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00021637506834335703,
+      "loss": 0.2569,
+      "step": 21250
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00021623838162930566,
+      "loss": 0.2673,
+      "step": 21260
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00021610169491525426,
+      "loss": 0.317,
+      "step": 21270
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00021596500820120286,
+      "loss": 0.2804,
+      "step": 21280
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00021582832148715146,
+      "loss": 0.2411,
+      "step": 21290
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00021569163477310006,
+      "loss": 0.2803,
+      "step": 21300
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00021555494805904866,
+      "loss": 0.2495,
+      "step": 21310
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00021541826134499726,
+      "loss": 0.2982,
+      "step": 21320
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00021528157463094588,
+      "loss": 0.2366,
+      "step": 21330
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00021514488791689448,
+      "loss": 0.2485,
+      "step": 21340
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00021500820120284308,
+      "loss": 0.2518,
+      "step": 21350
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00021487151448879168,
+      "loss": 0.2862,
+      "step": 21360
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00021473482777474028,
+      "loss": 0.259,
+      "step": 21370
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002145981410606889,
+      "loss": 0.2459,
+      "step": 21380
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00021446145434663753,
+      "loss": 0.2537,
+      "step": 21390
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00021432476763258613,
+      "loss": 0.2533,
+      "step": 21400
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00021418808091853473,
+      "loss": 0.2678,
+      "step": 21410
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00021405139420448333,
+      "loss": 0.2887,
+      "step": 21420
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00021391470749043193,
+      "loss": 0.2674,
+      "step": 21430
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00021377802077638053,
+      "loss": 0.2992,
+      "step": 21440
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00021364133406232913,
+      "loss": 0.2471,
+      "step": 21450
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00021350464734827776,
+      "loss": 0.2354,
+      "step": 21460
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00021336796063422636,
+      "loss": 0.2413,
+      "step": 21470
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00021323127392017496,
+      "loss": 0.2872,
+      "step": 21480
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00021309458720612356,
+      "loss": 0.2315,
+      "step": 21490
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00021295790049207216,
+      "loss": 0.2481,
+      "step": 21500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00021282121377802079,
+      "loss": 0.2347,
+      "step": 21510
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00021268452706396939,
+      "loss": 0.2745,
+      "step": 21520
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.000212547840349918,
+      "loss": 0.2953,
+      "step": 21530
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002124111536358666,
+      "loss": 0.2217,
+      "step": 21540
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002122744669218152,
+      "loss": 0.2006,
+      "step": 21550
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002121377802077638,
+      "loss": 0.3026,
+      "step": 21560
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002120010934937124,
+      "loss": 0.2621,
+      "step": 21570
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.000211864406779661,
+      "loss": 0.262,
+      "step": 21580
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002117277200656096,
+      "loss": 0.2507,
+      "step": 21590
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021159103335155824,
+      "loss": 0.2583,
+      "step": 21600
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021145434663750684,
+      "loss": 0.2884,
+      "step": 21610
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021131765992345544,
+      "loss": 0.2448,
+      "step": 21620
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021118097320940404,
+      "loss": 0.3132,
+      "step": 21630
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021104428649535264,
+      "loss": 0.3027,
+      "step": 21640
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021090759978130126,
+      "loss": 0.3007,
+      "step": 21650
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002107709130672499,
+      "loss": 0.2618,
+      "step": 21660
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002106342263531985,
+      "loss": 0.2551,
+      "step": 21670
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002104975396391471,
+      "loss": 0.2646,
+      "step": 21680
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002103608529250957,
+      "loss": 0.2478,
+      "step": 21690
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002102241662110443,
+      "loss": 0.3167,
+      "step": 21700
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002100874794969929,
+      "loss": 0.2895,
+      "step": 21710
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002099507927829415,
+      "loss": 0.3151,
+      "step": 21720
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020981410606889012,
+      "loss": 0.2807,
+      "step": 21730
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020967741935483871,
+      "loss": 0.2623,
+      "step": 21740
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020954073264078731,
+      "loss": 0.2486,
+      "step": 21750
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020940404592673591,
+      "loss": 0.2711,
+      "step": 21760
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020926735921268451,
+      "loss": 0.264,
+      "step": 21770
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020913067249863314,
+      "loss": 0.2543,
+      "step": 21780
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020899398578458174,
+      "loss": 0.2698,
+      "step": 21790
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020885729907053037,
+      "loss": 0.2494,
+      "step": 21800
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020872061235647897,
+      "loss": 0.2818,
+      "step": 21810
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020858392564242757,
+      "loss": 0.2308,
+      "step": 21820
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00020844723892837617,
+      "loss": 0.2726,
+      "step": 21830
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00020831055221432477,
+      "loss": 0.2786,
+      "step": 21840
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00020817386550027337,
+      "loss": 0.2976,
+      "step": 21850
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00020803717878622197,
+      "loss": 0.2519,
+      "step": 21860
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002079004920721706,
+      "loss": 0.3,
+      "step": 21870
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002077638053581192,
+      "loss": 0.2578,
+      "step": 21880
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002076271186440678,
+      "loss": 0.2828,
+      "step": 21890
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002074904319300164,
+      "loss": 0.2194,
+      "step": 21900
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020735374521596502,
+      "loss": 0.229,
+      "step": 21910
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020721705850191362,
+      "loss": 0.3025,
+      "step": 21920
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020708037178786225,
+      "loss": 0.2957,
+      "step": 21930
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020694368507381085,
+      "loss": 0.2562,
+      "step": 21940
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020680699835975944,
+      "loss": 0.2335,
+      "step": 21950
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020667031164570804,
+      "loss": 0.3255,
+      "step": 21960
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020653362493165664,
+      "loss": 0.2607,
+      "step": 21970
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020639693821760524,
+      "loss": 0.2257,
+      "step": 21980
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020626025150355384,
+      "loss": 0.253,
+      "step": 21990
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020612356478950247,
+      "loss": 0.2723,
+      "step": 22000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020598687807545107,
+      "loss": 0.3102,
+      "step": 22010
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020585019136139967,
+      "loss": 0.31,
+      "step": 22020
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020571350464734827,
+      "loss": 0.2503,
+      "step": 22030
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002055768179332969,
+      "loss": 0.2466,
+      "step": 22040
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002054401312192455,
+      "loss": 0.277,
+      "step": 22050
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002053034445051941,
+      "loss": 0.3027,
+      "step": 22060
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020516675779114272,
+      "loss": 0.2659,
+      "step": 22070
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020503007107709132,
+      "loss": 0.2086,
+      "step": 22080
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020489338436303992,
+      "loss": 0.2554,
+      "step": 22090
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020475669764898852,
+      "loss": 0.259,
+      "step": 22100
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020462001093493712,
+      "loss": 0.2784,
+      "step": 22110
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020448332422088572,
+      "loss": 0.2253,
+      "step": 22120
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020434663750683432,
+      "loss": 0.236,
+      "step": 22130
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020420995079278295,
+      "loss": 0.2906,
+      "step": 22140
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020407326407873155,
+      "loss": 0.2755,
+      "step": 22150
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020393657736468015,
+      "loss": 0.2887,
+      "step": 22160
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020379989065062877,
+      "loss": 0.2562,
+      "step": 22170
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020366320393657737,
+      "loss": 0.274,
+      "step": 22180
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020352651722252597,
+      "loss": 0.273,
+      "step": 22190
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002033898305084746,
+      "loss": 0.2886,
+      "step": 22200
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002032531437944232,
+      "loss": 0.2965,
+      "step": 22210
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002031164570803718,
+      "loss": 0.2516,
+      "step": 22220
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002029797703663204,
+      "loss": 0.2587,
+      "step": 22230
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.000202843083652269,
+      "loss": 0.2594,
+      "step": 22240
+    },
+    {
+      "epoch": 6.0,
+      "eval_Bleu_1": 0.004969491098949175,
+      "eval_Bleu_2": 2.062251367039122e-11,
+      "eval_Bleu_3": 3.667400956648284e-14,
+      "eval_Bleu_4": 1.6480142577061883e-15,
+      "eval_ROUGE_L": 0.007886930142300162,
+      "eval_cer": 0.5908475731463901,
+      "eval_em": 0.45109211775878444,
+      "eval_f1": 0.6069917591298192,
+      "eval_loss": 0.8702420592308044,
+      "eval_runtime": 312.0635,
+      "eval_samples_per_second": 13.497,
+      "eval_steps_per_second": 2.701,
+      "eval_wer": 0.6927360774818402,
+      "step": 22248
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002027063969382176,
+      "loss": 0.3088,
+      "step": 22250
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002025697102241662,
+      "loss": 0.1584,
+      "step": 22260
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020243302351011483,
+      "loss": 0.1669,
+      "step": 22270
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020229633679606343,
+      "loss": 0.1863,
+      "step": 22280
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020215965008201203,
+      "loss": 0.1528,
+      "step": 22290
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020202296336796063,
+      "loss": 0.1329,
+      "step": 22300
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00020188627665390925,
+      "loss": 0.1595,
+      "step": 22310
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00020174958993985785,
+      "loss": 0.1502,
+      "step": 22320
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00020161290322580645,
+      "loss": 0.1613,
+      "step": 22330
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00020147621651175508,
+      "loss": 0.1733,
+      "step": 22340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00020133952979770368,
+      "loss": 0.1654,
+      "step": 22350
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00020120284308365228,
+      "loss": 0.1323,
+      "step": 22360
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00020106615636960088,
+      "loss": 0.1373,
+      "step": 22370
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00020092946965554948,
+      "loss": 0.1868,
+      "step": 22380
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00020079278294149808,
+      "loss": 0.1465,
+      "step": 22390
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00020065609622744668,
+      "loss": 0.1447,
+      "step": 22400
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0002005194095133953,
+      "loss": 0.1292,
+      "step": 22410
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0002003827227993439,
+      "loss": 0.1412,
+      "step": 22420
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0002002460360852925,
+      "loss": 0.1878,
+      "step": 22430
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00020010934937124113,
+      "loss": 0.149,
+      "step": 22440
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019997266265718973,
+      "loss": 0.1449,
+      "step": 22450
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019983597594313833,
+      "loss": 0.1896,
+      "step": 22460
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019969928922908696,
+      "loss": 0.234,
+      "step": 22470
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019956260251503556,
+      "loss": 0.1743,
+      "step": 22480
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019942591580098416,
+      "loss": 0.2218,
+      "step": 22490
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019928922908693276,
+      "loss": 0.1709,
+      "step": 22500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019915254237288136,
+      "loss": 0.1641,
+      "step": 22510
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019901585565882995,
+      "loss": 0.2063,
+      "step": 22520
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019887916894477855,
+      "loss": 0.1918,
+      "step": 22530
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019874248223072718,
+      "loss": 0.2104,
+      "step": 22540
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019860579551667578,
+      "loss": 0.1368,
+      "step": 22550
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019846910880262438,
+      "loss": 0.1463,
+      "step": 22560
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.000198332422088573,
+      "loss": 0.1864,
+      "step": 22570
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001981957353745216,
+      "loss": 0.1958,
+      "step": 22580
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001980590486604702,
+      "loss": 0.175,
+      "step": 22590
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001979223619464188,
+      "loss": 0.1768,
+      "step": 22600
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019778567523236743,
+      "loss": 0.1414,
+      "step": 22610
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019764898851831603,
+      "loss": 0.2017,
+      "step": 22620
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019751230180426463,
+      "loss": 0.1581,
+      "step": 22630
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019737561509021323,
+      "loss": 0.1585,
+      "step": 22640
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019723892837616183,
+      "loss": 0.228,
+      "step": 22650
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019710224166211043,
+      "loss": 0.1781,
+      "step": 22660
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019696555494805903,
+      "loss": 0.1607,
+      "step": 22670
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019682886823400766,
+      "loss": 0.1666,
+      "step": 22680
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019669218151995626,
+      "loss": 0.1649,
+      "step": 22690
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019655549480590489,
+      "loss": 0.1836,
+      "step": 22700
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019641880809185349,
+      "loss": 0.1773,
+      "step": 22710
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019628212137780208,
+      "loss": 0.1451,
+      "step": 22720
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019614543466375068,
+      "loss": 0.1917,
+      "step": 22730
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001960087479496993,
+      "loss": 0.2318,
+      "step": 22740
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001958720612356479,
+      "loss": 0.1855,
+      "step": 22750
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001957353745215965,
+      "loss": 0.1621,
+      "step": 22760
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001955986878075451,
+      "loss": 0.182,
+      "step": 22770
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001954620010934937,
+      "loss": 0.2112,
+      "step": 22780
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001953253143794423,
+      "loss": 0.1843,
+      "step": 22790
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001951886276653909,
+      "loss": 0.1956,
+      "step": 22800
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019505194095133954,
+      "loss": 0.1505,
+      "step": 22810
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019491525423728814,
+      "loss": 0.1872,
+      "step": 22820
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019477856752323674,
+      "loss": 0.1797,
+      "step": 22830
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019464188080918536,
+      "loss": 0.1661,
+      "step": 22840
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019450519409513396,
+      "loss": 0.1745,
+      "step": 22850
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019436850738108256,
+      "loss": 0.1502,
+      "step": 22860
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019423182066703116,
+      "loss": 0.1272,
+      "step": 22870
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001940951339529798,
+      "loss": 0.162,
+      "step": 22880
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001939584472389284,
+      "loss": 0.1687,
+      "step": 22890
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000193821760524877,
+      "loss": 0.1509,
+      "step": 22900
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001936850738108256,
+      "loss": 0.2007,
+      "step": 22910
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001935483870967742,
+      "loss": 0.1947,
+      "step": 22920
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001934117003827228,
+      "loss": 0.1685,
+      "step": 22930
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001932750136686714,
+      "loss": 0.2043,
+      "step": 22940
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019313832695462001,
+      "loss": 0.1726,
+      "step": 22950
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019300164024056861,
+      "loss": 0.1919,
+      "step": 22960
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019286495352651724,
+      "loss": 0.1478,
+      "step": 22970
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019272826681246584,
+      "loss": 0.1733,
+      "step": 22980
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019259158009841444,
+      "loss": 0.164,
+      "step": 22990
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019245489338436304,
+      "loss": 0.1634,
+      "step": 23000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019231820667031167,
+      "loss": 0.1948,
+      "step": 23010
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019218151995626027,
+      "loss": 0.2086,
+      "step": 23020
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019204483324220887,
+      "loss": 0.1528,
+      "step": 23030
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019190814652815747,
+      "loss": 0.1777,
+      "step": 23040
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00019177145981410607,
+      "loss": 0.1557,
+      "step": 23050
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00019163477310005467,
+      "loss": 0.1405,
+      "step": 23060
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00019149808638600327,
+      "loss": 0.1753,
+      "step": 23070
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001913613996719519,
+      "loss": 0.1529,
+      "step": 23080
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001912247129579005,
+      "loss": 0.1623,
+      "step": 23090
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00019108802624384912,
+      "loss": 0.1586,
+      "step": 23100
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00019095133952979772,
+      "loss": 0.1392,
+      "step": 23110
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00019081465281574632,
+      "loss": 0.2238,
+      "step": 23120
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00019067796610169492,
+      "loss": 0.1772,
+      "step": 23130
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00019054127938764352,
+      "loss": 0.2431,
+      "step": 23140
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00019040459267359214,
+      "loss": 0.2054,
+      "step": 23150
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00019026790595954074,
+      "loss": 0.1816,
+      "step": 23160
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00019013121924548934,
+      "loss": 0.169,
+      "step": 23170
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018999453253143794,
+      "loss": 0.211,
+      "step": 23180
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018985784581738654,
+      "loss": 0.2138,
+      "step": 23190
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018972115910333514,
+      "loss": 0.1391,
+      "step": 23200
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018958447238928374,
+      "loss": 0.1656,
+      "step": 23210
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018944778567523237,
+      "loss": 0.1816,
+      "step": 23220
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.000189311098961181,
+      "loss": 0.2475,
+      "step": 23230
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001891744122471296,
+      "loss": 0.1454,
+      "step": 23240
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001890377255330782,
+      "loss": 0.1786,
+      "step": 23250
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001889010388190268,
+      "loss": 0.2279,
+      "step": 23260
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001887643521049754,
+      "loss": 0.1829,
+      "step": 23270
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018862766539092402,
+      "loss": 0.211,
+      "step": 23280
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018849097867687262,
+      "loss": 0.1682,
+      "step": 23290
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018835429196282122,
+      "loss": 0.1447,
+      "step": 23300
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018821760524876982,
+      "loss": 0.2046,
+      "step": 23310
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018808091853471842,
+      "loss": 0.2257,
+      "step": 23320
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018794423182066702,
+      "loss": 0.2074,
+      "step": 23330
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018780754510661562,
+      "loss": 0.1813,
+      "step": 23340
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018767085839256425,
+      "loss": 0.2057,
+      "step": 23350
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018753417167851285,
+      "loss": 0.223,
+      "step": 23360
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018739748496446147,
+      "loss": 0.1896,
+      "step": 23370
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018726079825041007,
+      "loss": 0.197,
+      "step": 23380
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018712411153635867,
+      "loss": 0.2221,
+      "step": 23390
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018698742482230727,
+      "loss": 0.1832,
+      "step": 23400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018685073810825587,
+      "loss": 0.1887,
+      "step": 23410
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001867140513942045,
+      "loss": 0.2102,
+      "step": 23420
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001865773646801531,
+      "loss": 0.179,
+      "step": 23430
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001864406779661017,
+      "loss": 0.177,
+      "step": 23440
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001863039912520503,
+      "loss": 0.1912,
+      "step": 23450
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001861673045379989,
+      "loss": 0.2105,
+      "step": 23460
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001860306178239475,
+      "loss": 0.196,
+      "step": 23470
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001858939311098961,
+      "loss": 0.1842,
+      "step": 23480
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00018575724439584473,
+      "loss": 0.194,
+      "step": 23490
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018562055768179335,
+      "loss": 0.1779,
+      "step": 23500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018548387096774195,
+      "loss": 0.1528,
+      "step": 23510
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018534718425369055,
+      "loss": 0.1728,
+      "step": 23520
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018521049753963915,
+      "loss": 0.2487,
+      "step": 23530
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018507381082558775,
+      "loss": 0.1393,
+      "step": 23540
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018493712411153638,
+      "loss": 0.2025,
+      "step": 23550
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018480043739748498,
+      "loss": 0.2122,
+      "step": 23560
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018466375068343358,
+      "loss": 0.1655,
+      "step": 23570
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018452706396938218,
+      "loss": 0.183,
+      "step": 23580
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018439037725533078,
+      "loss": 0.1803,
+      "step": 23590
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018425369054127938,
+      "loss": 0.1654,
+      "step": 23600
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018411700382722798,
+      "loss": 0.1524,
+      "step": 23610
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001839803171131766,
+      "loss": 0.1972,
+      "step": 23620
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018384363039912523,
+      "loss": 0.1634,
+      "step": 23630
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018370694368507383,
+      "loss": 0.1757,
+      "step": 23640
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018357025697102243,
+      "loss": 0.1934,
+      "step": 23650
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018343357025697103,
+      "loss": 0.1498,
+      "step": 23660
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018329688354291963,
+      "loss": 0.1836,
+      "step": 23670
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018316019682886823,
+      "loss": 0.179,
+      "step": 23680
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018302351011481686,
+      "loss": 0.179,
+      "step": 23690
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018288682340076545,
+      "loss": 0.1677,
+      "step": 23700
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018275013668671405,
+      "loss": 0.1917,
+      "step": 23710
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018261344997266265,
+      "loss": 0.1605,
+      "step": 23720
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018247676325861125,
+      "loss": 0.1924,
+      "step": 23730
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018234007654455985,
+      "loss": 0.1698,
+      "step": 23740
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00018220338983050845,
+      "loss": 0.2058,
+      "step": 23750
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001820667031164571,
+      "loss": 0.1788,
+      "step": 23760
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001819300164024057,
+      "loss": 0.1903,
+      "step": 23770
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001817933296883543,
+      "loss": 0.1423,
+      "step": 23780
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001816566429743029,
+      "loss": 0.2053,
+      "step": 23790
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001815199562602515,
+      "loss": 0.1914,
+      "step": 23800
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001813832695462001,
+      "loss": 0.1497,
+      "step": 23810
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018124658283214873,
+      "loss": 0.1671,
+      "step": 23820
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018110989611809733,
+      "loss": 0.1809,
+      "step": 23830
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018097320940404593,
+      "loss": 0.1647,
+      "step": 23840
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018083652268999453,
+      "loss": 0.1963,
+      "step": 23850
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018069983597594313,
+      "loss": 0.201,
+      "step": 23860
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018056314926189173,
+      "loss": 0.1727,
+      "step": 23870
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018042646254784033,
+      "loss": 0.1978,
+      "step": 23880
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018028977583378896,
+      "loss": 0.1944,
+      "step": 23890
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018015308911973759,
+      "loss": 0.1536,
+      "step": 23900
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018001640240568618,
+      "loss": 0.2317,
+      "step": 23910
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00017987971569163478,
+      "loss": 0.1597,
+      "step": 23920
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00017974302897758338,
+      "loss": 0.1966,
+      "step": 23930
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00017960634226353198,
+      "loss": 0.1609,
+      "step": 23940
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00017946965554948058,
+      "loss": 0.1555,
+      "step": 23950
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001793329688354292,
+      "loss": 0.1754,
+      "step": 23960
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001791962821213778,
+      "loss": 0.1698,
+      "step": 23970
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001790595954073264,
+      "loss": 0.1567,
+      "step": 23980
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.000178922908693275,
+      "loss": 0.2134,
+      "step": 23990
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001787862219792236,
+      "loss": 0.2524,
+      "step": 24000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001786495352651722,
+      "loss": 0.1614,
+      "step": 24010
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001785128485511208,
+      "loss": 0.1982,
+      "step": 24020
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00017837616183706946,
+      "loss": 0.211,
+      "step": 24030
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00017823947512301806,
+      "loss": 0.2004,
+      "step": 24040
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00017810278840896666,
+      "loss": 0.1624,
+      "step": 24050
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00017796610169491526,
+      "loss": 0.1678,
+      "step": 24060
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00017782941498086386,
+      "loss": 0.1988,
+      "step": 24070
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00017769272826681246,
+      "loss": 0.1845,
+      "step": 24080
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001775560415527611,
+      "loss": 0.2195,
+      "step": 24090
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001774193548387097,
+      "loss": 0.2059,
+      "step": 24100
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001772826681246583,
+      "loss": 0.1975,
+      "step": 24110
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001771459814106069,
+      "loss": 0.1809,
+      "step": 24120
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001770092946965555,
+      "loss": 0.1831,
+      "step": 24130
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001768726079825041,
+      "loss": 0.2218,
+      "step": 24140
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001767359212684527,
+      "loss": 0.2134,
+      "step": 24150
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00017659923455440134,
+      "loss": 0.1735,
+      "step": 24160
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00017646254784034994,
+      "loss": 0.2183,
+      "step": 24170
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00017632586112629854,
+      "loss": 0.1737,
+      "step": 24180
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00017618917441224714,
+      "loss": 0.2142,
+      "step": 24190
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00017605248769819574,
+      "loss": 0.1894,
+      "step": 24200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00017591580098414434,
+      "loss": 0.1819,
+      "step": 24210
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00017577911427009294,
+      "loss": 0.2059,
+      "step": 24220
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00017564242755604157,
+      "loss": 0.1709,
+      "step": 24230
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00017550574084199017,
+      "loss": 0.2048,
+      "step": 24240
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00017536905412793877,
+      "loss": 0.2548,
+      "step": 24250
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00017523236741388737,
+      "loss": 0.1541,
+      "step": 24260
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017509568069983596,
+      "loss": 0.1582,
+      "step": 24270
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017495899398578456,
+      "loss": 0.1698,
+      "step": 24280
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017482230727173316,
+      "loss": 0.1527,
+      "step": 24290
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017468562055768182,
+      "loss": 0.1973,
+      "step": 24300
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017454893384363042,
+      "loss": 0.1809,
+      "step": 24310
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017441224712957902,
+      "loss": 0.1636,
+      "step": 24320
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017427556041552762,
+      "loss": 0.1822,
+      "step": 24330
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017413887370147622,
+      "loss": 0.2203,
+      "step": 24340
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017400218698742482,
+      "loss": 0.1748,
+      "step": 24350
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017386550027337344,
+      "loss": 0.1975,
+      "step": 24360
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017372881355932204,
+      "loss": 0.2079,
+      "step": 24370
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017359212684527064,
+      "loss": 0.1923,
+      "step": 24380
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00017345544013121924,
+      "loss": 0.1941,
+      "step": 24390
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00017331875341716784,
+      "loss": 0.183,
+      "step": 24400
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00017318206670311644,
+      "loss": 0.1951,
+      "step": 24410
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00017304537998906504,
+      "loss": 0.1977,
+      "step": 24420
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001729086932750137,
+      "loss": 0.1957,
+      "step": 24430
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001727720065609623,
+      "loss": 0.2221,
+      "step": 24440
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001726353198469109,
+      "loss": 0.159,
+      "step": 24450
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001724986331328595,
+      "loss": 0.2052,
+      "step": 24460
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001723619464188081,
+      "loss": 0.2274,
+      "step": 24470
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001722252597047567,
+      "loss": 0.2071,
+      "step": 24480
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001720885729907053,
+      "loss": 0.2199,
+      "step": 24490
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017195188627665392,
+      "loss": 0.1498,
+      "step": 24500
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017181519956260252,
+      "loss": 0.1988,
+      "step": 24510
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017167851284855112,
+      "loss": 0.2155,
+      "step": 24520
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017154182613449972,
+      "loss": 0.1821,
+      "step": 24530
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017140513942044832,
+      "loss": 0.183,
+      "step": 24540
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017126845270639692,
+      "loss": 0.2257,
+      "step": 24550
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017113176599234555,
+      "loss": 0.2076,
+      "step": 24560
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017099507927829417,
+      "loss": 0.1692,
+      "step": 24570
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017085839256424277,
+      "loss": 0.1857,
+      "step": 24580
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017072170585019137,
+      "loss": 0.1877,
+      "step": 24590
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017058501913613997,
+      "loss": 0.2081,
+      "step": 24600
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017044833242208857,
+      "loss": 0.1709,
+      "step": 24610
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017031164570803717,
+      "loss": 0.2158,
+      "step": 24620
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001701749589939858,
+      "loss": 0.2011,
+      "step": 24630
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001700382722799344,
+      "loss": 0.2356,
+      "step": 24640
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.000169901585565883,
+      "loss": 0.201,
+      "step": 24650
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001697648988518316,
+      "loss": 0.1814,
+      "step": 24660
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001696282121377802,
+      "loss": 0.1828,
+      "step": 24670
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001694915254237288,
+      "loss": 0.2306,
+      "step": 24680
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00016935483870967742,
+      "loss": 0.1594,
+      "step": 24690
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00016921815199562605,
+      "loss": 0.1561,
+      "step": 24700
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00016908146528157465,
+      "loss": 0.2069,
+      "step": 24710
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00016894477856752325,
+      "loss": 0.2014,
+      "step": 24720
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00016880809185347185,
+      "loss": 0.1876,
+      "step": 24730
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00016867140513942045,
+      "loss": 0.1844,
+      "step": 24740
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00016853471842536905,
+      "loss": 0.2265,
+      "step": 24750
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00016839803171131765,
+      "loss": 0.1851,
+      "step": 24760
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00016826134499726628,
+      "loss": 0.18,
+      "step": 24770
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00016812465828321488,
+      "loss": 0.2304,
+      "step": 24780
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00016798797156916348,
+      "loss": 0.2079,
+      "step": 24790
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00016785128485511208,
+      "loss": 0.2318,
+      "step": 24800
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00016771459814106068,
+      "loss": 0.1934,
+      "step": 24810
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00016757791142700928,
+      "loss": 0.1705,
+      "step": 24820
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001674412247129579,
+      "loss": 0.1914,
+      "step": 24830
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00016730453799890653,
+      "loss": 0.1966,
+      "step": 24840
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00016716785128485513,
+      "loss": 0.1488,
+      "step": 24850
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00016703116457080373,
+      "loss": 0.1592,
+      "step": 24860
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00016689447785675233,
+      "loss": 0.2195,
+      "step": 24870
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00016675779114270093,
+      "loss": 0.2083,
+      "step": 24880
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00016662110442864953,
+      "loss": 0.1671,
+      "step": 24890
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00016648441771459815,
+      "loss": 0.1977,
+      "step": 24900
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00016634773100054675,
+      "loss": 0.1761,
+      "step": 24910
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00016621104428649535,
+      "loss": 0.2144,
+      "step": 24920
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00016607435757244395,
+      "loss": 0.2269,
+      "step": 24930
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00016593767085839255,
+      "loss": 0.2128,
+      "step": 24940
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00016580098414434115,
+      "loss": 0.1856,
+      "step": 24950
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00016566429743028978,
+      "loss": 0.1598,
+      "step": 24960
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001655276107162384,
+      "loss": 0.2002,
+      "step": 24970
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.000165390924002187,
+      "loss": 0.1768,
+      "step": 24980
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001652542372881356,
+      "loss": 0.2017,
+      "step": 24990
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001651175505740842,
+      "loss": 0.171,
+      "step": 25000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001649808638600328,
+      "loss": 0.1638,
+      "step": 25010
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001648441771459814,
+      "loss": 0.2331,
+      "step": 25020
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00016470749043193,
+      "loss": 0.1699,
+      "step": 25030
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00016457080371787863,
+      "loss": 0.2023,
+      "step": 25040
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00016443411700382723,
+      "loss": 0.1865,
+      "step": 25050
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00016429743028977583,
+      "loss": 0.1952,
+      "step": 25060
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00016416074357572443,
+      "loss": 0.1885,
+      "step": 25070
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00016402405686167303,
+      "loss": 0.2217,
+      "step": 25080
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00016388737014762166,
+      "loss": 0.1994,
+      "step": 25090
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00016375068343357026,
+      "loss": 0.1743,
+      "step": 25100
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00016361399671951888,
+      "loss": 0.1948,
+      "step": 25110
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00016347731000546748,
+      "loss": 0.2518,
+      "step": 25120
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00016334062329141608,
+      "loss": 0.1976,
+      "step": 25130
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00016320393657736468,
+      "loss": 0.2284,
+      "step": 25140
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00016306724986331328,
+      "loss": 0.1743,
+      "step": 25150
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00016293056314926188,
+      "loss": 0.2376,
+      "step": 25160
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001627938764352105,
+      "loss": 0.2335,
+      "step": 25170
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001626571897211591,
+      "loss": 0.221,
+      "step": 25180
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001625205030071077,
+      "loss": 0.1576,
+      "step": 25190
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001623838162930563,
+      "loss": 0.2081,
+      "step": 25200
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001622471295790049,
+      "loss": 0.1645,
+      "step": 25210
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00016211044286495354,
+      "loss": 0.2039,
+      "step": 25220
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00016197375615090214,
+      "loss": 0.1774,
+      "step": 25230
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00016183706943685076,
+      "loss": 0.233,
+      "step": 25240
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00016170038272279936,
+      "loss": 0.2315,
+      "step": 25250
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00016156369600874796,
+      "loss": 0.1958,
+      "step": 25260
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00016142700929469656,
+      "loss": 0.1422,
+      "step": 25270
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00016129032258064516,
+      "loss": 0.2104,
+      "step": 25280
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00016115363586659376,
+      "loss": 0.169,
+      "step": 25290
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00016101694915254236,
+      "loss": 0.2042,
+      "step": 25300
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.000160880262438491,
+      "loss": 0.1755,
+      "step": 25310
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001607435757244396,
+      "loss": 0.1992,
+      "step": 25320
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001606068890103882,
+      "loss": 0.1984,
+      "step": 25330
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001604702022963368,
+      "loss": 0.1453,
+      "step": 25340
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00016033351558228539,
+      "loss": 0.2067,
+      "step": 25350
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.000160196828868234,
+      "loss": 0.171,
+      "step": 25360
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001600601421541826,
+      "loss": 0.1744,
+      "step": 25370
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00015992345544013124,
+      "loss": 0.1979,
+      "step": 25380
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00015978676872607984,
+      "loss": 0.2077,
+      "step": 25390
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00015965008201202844,
+      "loss": 0.2215,
+      "step": 25400
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00015951339529797704,
+      "loss": 0.2019,
+      "step": 25410
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00015937670858392564,
+      "loss": 0.1845,
+      "step": 25420
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00015924002186987424,
+      "loss": 0.2363,
+      "step": 25430
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00015910333515582287,
+      "loss": 0.1516,
+      "step": 25440
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00015896664844177147,
+      "loss": 0.1904,
+      "step": 25450
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00015882996172772006,
+      "loss": 0.2374,
+      "step": 25460
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00015869327501366866,
+      "loss": 0.1625,
+      "step": 25470
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00015855658829961726,
+      "loss": 0.186,
+      "step": 25480
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001584199015855659,
+      "loss": 0.1802,
+      "step": 25490
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001582832148715145,
+      "loss": 0.2023,
+      "step": 25500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00015814652815746312,
+      "loss": 0.2049,
+      "step": 25510
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00015800984144341172,
+      "loss": 0.1627,
+      "step": 25520
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00015787315472936032,
+      "loss": 0.1944,
+      "step": 25530
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00015773646801530892,
+      "loss": 0.1933,
+      "step": 25540
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00015759978130125752,
+      "loss": 0.2297,
+      "step": 25550
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00015746309458720612,
+      "loss": 0.1727,
+      "step": 25560
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00015732640787315472,
+      "loss": 0.1453,
+      "step": 25570
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00015718972115910334,
+      "loss": 0.1677,
+      "step": 25580
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00015705303444505194,
+      "loss": 0.1793,
+      "step": 25590
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00015691634773100054,
+      "loss": 0.1483,
+      "step": 25600
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00015677966101694914,
+      "loss": 0.2071,
+      "step": 25610
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00015664297430289777,
+      "loss": 0.1847,
+      "step": 25620
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00015650628758884637,
+      "loss": 0.1785,
+      "step": 25630
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00015636960087479497,
+      "loss": 0.1894,
+      "step": 25640
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001562329141607436,
+      "loss": 0.2217,
+      "step": 25650
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001560962274466922,
+      "loss": 0.1955,
+      "step": 25660
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001559595407326408,
+      "loss": 0.201,
+      "step": 25670
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001558228540185894,
+      "loss": 0.194,
+      "step": 25680
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.000155686167304538,
+      "loss": 0.21,
+      "step": 25690
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001555494805904866,
+      "loss": 0.2125,
+      "step": 25700
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00015541279387643522,
+      "loss": 0.2056,
+      "step": 25710
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00015527610716238382,
+      "loss": 0.2238,
+      "step": 25720
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00015513942044833242,
+      "loss": 0.1665,
+      "step": 25730
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00015500273373428102,
+      "loss": 0.2267,
+      "step": 25740
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00015486604702022965,
+      "loss": 0.2093,
+      "step": 25750
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00015472936030617825,
+      "loss": 0.1941,
+      "step": 25760
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00015459267359212685,
+      "loss": 0.1726,
+      "step": 25770
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00015445598687807547,
+      "loss": 0.2079,
+      "step": 25780
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00015431930016402407,
+      "loss": 0.1742,
+      "step": 25790
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00015418261344997267,
+      "loss": 0.1807,
+      "step": 25800
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00015404592673592127,
+      "loss": 0.2019,
+      "step": 25810
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00015390924002186987,
+      "loss": 0.191,
+      "step": 25820
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00015377255330781847,
+      "loss": 0.2067,
+      "step": 25830
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00015363586659376707,
+      "loss": 0.1783,
+      "step": 25840
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001534991798797157,
+      "loss": 0.2038,
+      "step": 25850
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001533624931656643,
+      "loss": 0.1807,
+      "step": 25860
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0001532258064516129,
+      "loss": 0.2086,
+      "step": 25870
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0001530891197375615,
+      "loss": 0.2166,
+      "step": 25880
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00015295243302351012,
+      "loss": 0.2433,
+      "step": 25890
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00015281574630945872,
+      "loss": 0.1749,
+      "step": 25900
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00015267905959540732,
+      "loss": 0.1711,
+      "step": 25910
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00015254237288135595,
+      "loss": 0.2248,
+      "step": 25920
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00015240568616730455,
+      "loss": 0.1933,
+      "step": 25930
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00015226899945325315,
+      "loss": 0.2053,
+      "step": 25940
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00015213231273920175,
+      "loss": 0.1674,
+      "step": 25950
+    },
+    {
+      "epoch": 7.0,
+      "eval_Bleu_1": 0.004898218829516228,
+      "eval_Bleu_2": 2.0630947082931953e-11,
+      "eval_Bleu_3": 3.689488897487006e-14,
+      "eval_Bleu_4": 1.661511311162505e-15,
+      "eval_ROUGE_L": 0.007489217110335963,
+      "eval_cer": 0.585765055255343,
+      "eval_em": 0.4601139601139601,
+      "eval_f1": 0.6149240917751063,
+      "eval_loss": 0.893440842628479,
+      "eval_runtime": 313.8376,
+      "eval_samples_per_second": 13.421,
+      "eval_steps_per_second": 2.686,
+      "eval_wer": 0.6889225181598063,
+      "step": 25956
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00015199562602515035,
+      "loss": 0.1353,
+      "step": 25960
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00015185893931109895,
+      "loss": 0.0992,
+      "step": 25970
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00015172225259704758,
+      "loss": 0.1154,
+      "step": 25980
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00015158556588299618,
+      "loss": 0.1172,
+      "step": 25990
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00015144887916894478,
+      "loss": 0.1525,
+      "step": 26000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00015131219245489338,
+      "loss": 0.1388,
+      "step": 26010
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.000151175505740842,
+      "loss": 0.141,
+      "step": 26020
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0001510388190267906,
+      "loss": 0.1141,
+      "step": 26030
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0001509021323127392,
+      "loss": 0.1346,
+      "step": 26040
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00015076544559868783,
+      "loss": 0.1392,
+      "step": 26050
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00015062875888463643,
+      "loss": 0.1311,
+      "step": 26060
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00015049207217058503,
+      "loss": 0.1279,
+      "step": 26070
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00015035538545653363,
+      "loss": 0.1373,
+      "step": 26080
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00015021869874248223,
+      "loss": 0.1042,
+      "step": 26090
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00015008201202843083,
+      "loss": 0.1286,
+      "step": 26100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00014994532531437943,
+      "loss": 0.1013,
+      "step": 26110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00014980863860032805,
+      "loss": 0.0924,
+      "step": 26120
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00014967195188627665,
+      "loss": 0.11,
+      "step": 26130
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00014953526517222525,
+      "loss": 0.1189,
+      "step": 26140
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00014939857845817388,
+      "loss": 0.1375,
+      "step": 26150
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00014926189174412248,
+      "loss": 0.1172,
+      "step": 26160
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00014912520503007108,
+      "loss": 0.1041,
+      "step": 26170
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00014898851831601968,
+      "loss": 0.1104,
+      "step": 26180
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0001488518316019683,
+      "loss": 0.0947,
+      "step": 26190
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001487151448879169,
+      "loss": 0.1457,
+      "step": 26200
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001485784581738655,
+      "loss": 0.1544,
+      "step": 26210
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001484417714598141,
+      "loss": 0.123,
+      "step": 26220
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001483050847457627,
+      "loss": 0.1278,
+      "step": 26230
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0001481683980317113,
+      "loss": 0.1113,
+      "step": 26240
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00014803171131765993,
+      "loss": 0.1272,
+      "step": 26250
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00014789502460360853,
+      "loss": 0.1305,
+      "step": 26260
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00014775833788955713,
+      "loss": 0.1366,
+      "step": 26270
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00014762165117550576,
+      "loss": 0.1066,
+      "step": 26280
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00014748496446145436,
+      "loss": 0.1248,
+      "step": 26290
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00014734827774740296,
+      "loss": 0.1526,
+      "step": 26300
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00014721159103335156,
+      "loss": 0.1268,
+      "step": 26310
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00014707490431930018,
+      "loss": 0.1683,
+      "step": 26320
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00014693821760524878,
+      "loss": 0.115,
+      "step": 26330
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00014680153089119738,
+      "loss": 0.1315,
+      "step": 26340
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00014666484417714598,
+      "loss": 0.1283,
+      "step": 26350
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00014652815746309458,
+      "loss": 0.1437,
+      "step": 26360
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00014639147074904318,
+      "loss": 0.1102,
+      "step": 26370
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00014625478403499178,
+      "loss": 0.1621,
+      "step": 26380
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0001461180973209404,
+      "loss": 0.1014,
+      "step": 26390
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.000145981410606889,
+      "loss": 0.1595,
+      "step": 26400
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0001458447238928376,
+      "loss": 0.1394,
+      "step": 26410
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00014570803717878624,
+      "loss": 0.1113,
+      "step": 26420
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00014557135046473483,
+      "loss": 0.1292,
+      "step": 26430
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00014543466375068343,
+      "loss": 0.1363,
+      "step": 26440
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00014529797703663203,
+      "loss": 0.1308,
+      "step": 26450
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00014516129032258066,
+      "loss": 0.1132,
+      "step": 26460
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00014502460360852926,
+      "loss": 0.1295,
+      "step": 26470
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00014488791689447786,
+      "loss": 0.1647,
+      "step": 26480
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00014475123018042646,
+      "loss": 0.1212,
+      "step": 26490
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00014461454346637506,
+      "loss": 0.1199,
+      "step": 26500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00014447785675232366,
+      "loss": 0.1533,
+      "step": 26510
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0001443411700382723,
+      "loss": 0.1321,
+      "step": 26520
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0001442044833242209,
+      "loss": 0.1435,
+      "step": 26530
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00014406779661016949,
+      "loss": 0.1131,
+      "step": 26540
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0001439311098961181,
+      "loss": 0.1064,
+      "step": 26550
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0001437944231820667,
+      "loss": 0.149,
+      "step": 26560
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0001436577364680153,
+      "loss": 0.1533,
+      "step": 26570
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0001435210497539639,
+      "loss": 0.1188,
+      "step": 26580
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00014338436303991254,
+      "loss": 0.1533,
+      "step": 26590
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00014324767632586114,
+      "loss": 0.1236,
+      "step": 26600
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00014311098961180974,
+      "loss": 0.1083,
+      "step": 26610
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00014297430289775834,
+      "loss": 0.1428,
+      "step": 26620
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00014283761618370694,
+      "loss": 0.1435,
+      "step": 26630
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00014270092946965554,
+      "loss": 0.1264,
+      "step": 26640
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00014256424275560414,
+      "loss": 0.1363,
+      "step": 26650
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00014242755604155276,
+      "loss": 0.1447,
+      "step": 26660
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00014229086932750136,
+      "loss": 0.1294,
+      "step": 26670
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00014215418261345,
+      "loss": 0.1539,
+      "step": 26680
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001420174958993986,
+      "loss": 0.1408,
+      "step": 26690
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001418808091853472,
+      "loss": 0.1601,
+      "step": 26700
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001417441224712958,
+      "loss": 0.1512,
+      "step": 26710
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0001416074357572444,
+      "loss": 0.1421,
+      "step": 26720
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00014147074904319302,
+      "loss": 0.1068,
+      "step": 26730
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00014133406232914162,
+      "loss": 0.1784,
+      "step": 26740
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00014119737561509022,
+      "loss": 0.153,
+      "step": 26750
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00014106068890103882,
+      "loss": 0.1444,
+      "step": 26760
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00014092400218698742,
+      "loss": 0.1106,
+      "step": 26770
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00014078731547293602,
+      "loss": 0.1261,
+      "step": 26780
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00014065062875888464,
+      "loss": 0.1491,
+      "step": 26790
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00014051394204483324,
+      "loss": 0.1182,
+      "step": 26800
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00014037725533078187,
+      "loss": 0.0927,
+      "step": 26810
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00014024056861673047,
+      "loss": 0.1275,
+      "step": 26820
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00014010388190267907,
+      "loss": 0.1431,
+      "step": 26830
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00013996719518862767,
+      "loss": 0.1299,
+      "step": 26840
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00013983050847457627,
+      "loss": 0.1316,
+      "step": 26850
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0001396938217605249,
+      "loss": 0.1433,
+      "step": 26860
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0001395571350464735,
+      "loss": 0.147,
+      "step": 26870
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0001394204483324221,
+      "loss": 0.1205,
+      "step": 26880
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0001392837616183707,
+      "loss": 0.1397,
+      "step": 26890
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0001391470749043193,
+      "loss": 0.1516,
+      "step": 26900
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0001390103881902679,
+      "loss": 0.1294,
+      "step": 26910
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0001388737014762165,
+      "loss": 0.127,
+      "step": 26920
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00013873701476216512,
+      "loss": 0.112,
+      "step": 26930
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00013860032804811372,
+      "loss": 0.1435,
+      "step": 26940
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00013846364133406235,
+      "loss": 0.1543,
+      "step": 26950
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00013832695462001095,
+      "loss": 0.1162,
+      "step": 26960
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00013819026790595955,
+      "loss": 0.1111,
+      "step": 26970
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00013805358119190815,
+      "loss": 0.132,
+      "step": 26980
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00013791689447785675,
+      "loss": 0.1184,
+      "step": 26990
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00013778020776380537,
+      "loss": 0.1043,
+      "step": 27000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00013764352104975397,
+      "loss": 0.1635,
+      "step": 27010
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00013750683433570257,
+      "loss": 0.1245,
+      "step": 27020
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00013737014762165117,
+      "loss": 0.1411,
+      "step": 27030
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00013723346090759977,
+      "loss": 0.1314,
+      "step": 27040
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00013709677419354837,
+      "loss": 0.1562,
+      "step": 27050
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.000136960087479497,
+      "loss": 0.1061,
+      "step": 27060
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0001368234007654456,
+      "loss": 0.0837,
+      "step": 27070
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00013668671405139422,
+      "loss": 0.1057,
+      "step": 27080
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00013655002733734282,
+      "loss": 0.1093,
+      "step": 27090
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00013641334062329142,
+      "loss": 0.117,
+      "step": 27100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00013627665390924002,
+      "loss": 0.1273,
+      "step": 27110
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00013613996719518862,
+      "loss": 0.1428,
+      "step": 27120
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00013600328048113725,
+      "loss": 0.0991,
+      "step": 27130
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00013586659376708585,
+      "loss": 0.1322,
+      "step": 27140
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00013572990705303445,
+      "loss": 0.1377,
+      "step": 27150
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00013559322033898305,
+      "loss": 0.1424,
+      "step": 27160
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00013545653362493165,
+      "loss": 0.1084,
+      "step": 27170
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00013531984691088025,
+      "loss": 0.1581,
+      "step": 27180
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00013518316019682885,
+      "loss": 0.1572,
+      "step": 27190
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00013504647348277748,
+      "loss": 0.1304,
+      "step": 27200
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0001349097867687261,
+      "loss": 0.1343,
+      "step": 27210
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0001347731000546747,
+      "loss": 0.129,
+      "step": 27220
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0001346364133406233,
+      "loss": 0.153,
+      "step": 27230
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0001344997266265719,
+      "loss": 0.1462,
+      "step": 27240
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0001343630399125205,
+      "loss": 0.1599,
+      "step": 27250
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0001342263531984691,
+      "loss": 0.1533,
+      "step": 27260
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00013408966648441773,
+      "loss": 0.1462,
+      "step": 27270
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00013395297977036633,
+      "loss": 0.1447,
+      "step": 27280
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00013381629305631493,
+      "loss": 0.1232,
+      "step": 27290
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00013367960634226353,
+      "loss": 0.146,
+      "step": 27300
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00013354291962821213,
+      "loss": 0.159,
+      "step": 27310
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00013340623291416073,
+      "loss": 0.1524,
+      "step": 27320
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00013326954620010935,
+      "loss": 0.1254,
+      "step": 27330
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00013313285948605798,
+      "loss": 0.1371,
+      "step": 27340
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00013299617277200658,
+      "loss": 0.1422,
+      "step": 27350
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00013285948605795518,
+      "loss": 0.1466,
+      "step": 27360
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00013272279934390378,
+      "loss": 0.1219,
+      "step": 27370
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00013258611262985238,
+      "loss": 0.1407,
+      "step": 27380
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00013244942591580098,
+      "loss": 0.1343,
+      "step": 27390
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001323127392017496,
+      "loss": 0.157,
+      "step": 27400
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001321760524876982,
+      "loss": 0.1775,
+      "step": 27410
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001320393657736468,
+      "loss": 0.1511,
+      "step": 27420
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0001319026790595954,
+      "loss": 0.1733,
+      "step": 27430
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.000131765992345544,
+      "loss": 0.1395,
+      "step": 27440
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0001316293056314926,
+      "loss": 0.1173,
+      "step": 27450
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001314926189174412,
+      "loss": 0.1387,
+      "step": 27460
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00013135593220338986,
+      "loss": 0.1059,
+      "step": 27470
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00013121924548933846,
+      "loss": 0.1519,
+      "step": 27480
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00013108255877528706,
+      "loss": 0.1319,
+      "step": 27490
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00013094587206123566,
+      "loss": 0.1116,
+      "step": 27500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00013080918534718426,
+      "loss": 0.1658,
+      "step": 27510
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00013067249863313286,
+      "loss": 0.1274,
+      "step": 27520
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00013053581191908146,
+      "loss": 0.094,
+      "step": 27530
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00013039912520503008,
+      "loss": 0.1265,
+      "step": 27540
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00013026243849097868,
+      "loss": 0.1388,
+      "step": 27550
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00013012575177692728,
+      "loss": 0.1111,
+      "step": 27560
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00012998906506287588,
+      "loss": 0.1292,
+      "step": 27570
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00012985237834882448,
+      "loss": 0.1367,
+      "step": 27580
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00012971569163477308,
+      "loss": 0.1564,
+      "step": 27590
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0001295790049207217,
+      "loss": 0.1301,
+      "step": 27600
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00012944231820667034,
+      "loss": 0.1172,
+      "step": 27610
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00012930563149261893,
+      "loss": 0.1276,
+      "step": 27620
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00012916894477856753,
+      "loss": 0.1235,
+      "step": 27630
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00012903225806451613,
+      "loss": 0.1373,
+      "step": 27640
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00012889557135046473,
+      "loss": 0.1168,
+      "step": 27650
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00012875888463641333,
+      "loss": 0.1465,
+      "step": 27660
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00012862219792236196,
+      "loss": 0.1252,
+      "step": 27670
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00012848551120831056,
+      "loss": 0.1404,
+      "step": 27680
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00012834882449425916,
+      "loss": 0.1147,
+      "step": 27690
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00012821213778020776,
+      "loss": 0.1214,
+      "step": 27700
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00012807545106615636,
+      "loss": 0.134,
+      "step": 27710
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00012793876435210496,
+      "loss": 0.1438,
+      "step": 27720
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00012780207763805356,
+      "loss": 0.1256,
+      "step": 27730
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0001276653909240022,
+      "loss": 0.1261,
+      "step": 27740
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0001275287042099508,
+      "loss": 0.143,
+      "step": 27750
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001273920174958994,
+      "loss": 0.132,
+      "step": 27760
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.000127255330781848,
+      "loss": 0.1459,
+      "step": 27770
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001271186440677966,
+      "loss": 0.1403,
+      "step": 27780
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001269819573537452,
+      "loss": 0.1379,
+      "step": 27790
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0001268452706396938,
+      "loss": 0.1265,
+      "step": 27800
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00012670858392564244,
+      "loss": 0.1687,
+      "step": 27810
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00012657189721159104,
+      "loss": 0.1734,
+      "step": 27820
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00012643521049753964,
+      "loss": 0.125,
+      "step": 27830
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00012629852378348824,
+      "loss": 0.1229,
+      "step": 27840
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00012616183706943684,
+      "loss": 0.1656,
+      "step": 27850
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00012602515035538544,
+      "loss": 0.1698,
+      "step": 27860
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001258884636413341,
+      "loss": 0.1152,
+      "step": 27870
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001257517769272827,
+      "loss": 0.1367,
+      "step": 27880
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001256150902132313,
+      "loss": 0.156,
+      "step": 27890
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001254784034991799,
+      "loss": 0.1473,
+      "step": 27900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0001253417167851285,
+      "loss": 0.1078,
+      "step": 27910
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0001252050300710771,
+      "loss": 0.1306,
+      "step": 27920
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0001250683433570257,
+      "loss": 0.1486,
+      "step": 27930
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00012493165664297432,
+      "loss": 0.1067,
+      "step": 27940
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00012479496992892292,
+      "loss": 0.1099,
+      "step": 27950
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00012465828321487152,
+      "loss": 0.1603,
+      "step": 27960
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00012452159650082012,
+      "loss": 0.1355,
+      "step": 27970
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00012438490978676871,
+      "loss": 0.1433,
+      "step": 27980
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00012424822307271734,
+      "loss": 0.1112,
+      "step": 27990
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00012411153635866594,
+      "loss": 0.1166,
+      "step": 28000
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00012397484964461454,
+      "loss": 0.1658,
+      "step": 28010
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00012383816293056317,
+      "loss": 0.1469,
+      "step": 28020
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00012370147621651177,
+      "loss": 0.1489,
+      "step": 28030
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00012356478950246037,
+      "loss": 0.1389,
+      "step": 28040
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00012342810278840897,
+      "loss": 0.1718,
+      "step": 28050
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.00012329141607435757,
+      "loss": 0.1434,
+      "step": 28060
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0001231547293603062,
+      "loss": 0.1377,
+      "step": 28070
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0001230180426462548,
+      "loss": 0.1188,
+      "step": 28080
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0001228813559322034,
+      "loss": 0.142,
+      "step": 28090
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.000122744669218152,
+      "loss": 0.1325,
+      "step": 28100
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0001226079825041006,
+      "loss": 0.1167,
+      "step": 28110
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0001224712957900492,
+      "loss": 0.1638,
+      "step": 28120
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00012233460907599782,
+      "loss": 0.1236,
+      "step": 28130
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00012219792236194642,
+      "loss": 0.1252,
+      "step": 28140
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00012206123564789502,
+      "loss": 0.1192,
+      "step": 28150
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00012192454893384363,
+      "loss": 0.1478,
+      "step": 28160
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00012178786221979225,
+      "loss": 0.1442,
+      "step": 28170
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00012165117550574085,
+      "loss": 0.1539,
+      "step": 28180
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00012151448879168946,
+      "loss": 0.1249,
+      "step": 28190
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00012137780207763806,
+      "loss": 0.1785,
+      "step": 28200
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00012124111536358666,
+      "loss": 0.1426,
+      "step": 28210
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00012110442864953526,
+      "loss": 0.1356,
+      "step": 28220
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00012096774193548387,
+      "loss": 0.151,
+      "step": 28230
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00012083105522143248,
+      "loss": 0.1575,
+      "step": 28240
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00012069436850738108,
+      "loss": 0.1699,
+      "step": 28250
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0001205576817933297,
+      "loss": 0.1324,
+      "step": 28260
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0001204209950792783,
+      "loss": 0.1226,
+      "step": 28270
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.0001202843083652269,
+      "loss": 0.1189,
+      "step": 28280
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00012014762165117551,
+      "loss": 0.1553,
+      "step": 28290
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00012001093493712412,
+      "loss": 0.1397,
+      "step": 28300
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00011987424822307272,
+      "loss": 0.114,
+      "step": 28310
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00011973756150902132,
+      "loss": 0.1177,
+      "step": 28320
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00011960087479496994,
+      "loss": 0.166,
+      "step": 28330
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00011946418808091854,
+      "loss": 0.1062,
+      "step": 28340
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00011932750136686714,
+      "loss": 0.1354,
+      "step": 28350
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00011919081465281575,
+      "loss": 0.1066,
+      "step": 28360
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00011905412793876436,
+      "loss": 0.1343,
+      "step": 28370
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00011891744122471296,
+      "loss": 0.1228,
+      "step": 28380
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00011878075451066156,
+      "loss": 0.1446,
+      "step": 28390
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00011864406779661017,
+      "loss": 0.1905,
+      "step": 28400
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00011850738108255877,
+      "loss": 0.173,
+      "step": 28410
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00011837069436850737,
+      "loss": 0.1536,
+      "step": 28420
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.00011823400765445599,
+      "loss": 0.1377,
+      "step": 28430
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0001180973209404046,
+      "loss": 0.1276,
+      "step": 28440
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0001179606342263532,
+      "loss": 0.1088,
+      "step": 28450
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00011782394751230181,
+      "loss": 0.1792,
+      "step": 28460
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00011768726079825041,
+      "loss": 0.1398,
+      "step": 28470
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00011755057408419901,
+      "loss": 0.1518,
+      "step": 28480
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00011741388737014761,
+      "loss": 0.1548,
+      "step": 28490
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00011727720065609624,
+      "loss": 0.1393,
+      "step": 28500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00011714051394204484,
+      "loss": 0.148,
+      "step": 28510
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00011700382722799344,
+      "loss": 0.1345,
+      "step": 28520
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00011686714051394205,
+      "loss": 0.1139,
+      "step": 28530
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00011673045379989065,
+      "loss": 0.1289,
+      "step": 28540
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00011659376708583925,
+      "loss": 0.1713,
+      "step": 28550
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00011645708037178787,
+      "loss": 0.1722,
+      "step": 28560
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00011632039365773648,
+      "loss": 0.1209,
+      "step": 28570
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00011618370694368508,
+      "loss": 0.1546,
+      "step": 28580
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00011604702022963368,
+      "loss": 0.1425,
+      "step": 28590
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00011591033351558229,
+      "loss": 0.151,
+      "step": 28600
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00011577364680153089,
+      "loss": 0.1434,
+      "step": 28610
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00011563696008747949,
+      "loss": 0.1486,
+      "step": 28620
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0001155002733734281,
+      "loss": 0.1552,
+      "step": 28630
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00011536358665937672,
+      "loss": 0.1045,
+      "step": 28640
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00011522689994532532,
+      "loss": 0.1601,
+      "step": 28650
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00011509021323127392,
+      "loss": 0.114,
+      "step": 28660
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00011495352651722253,
+      "loss": 0.1418,
+      "step": 28670
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00011481683980317113,
+      "loss": 0.1798,
+      "step": 28680
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00011468015308911973,
+      "loss": 0.1318,
+      "step": 28690
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00011454346637506836,
+      "loss": 0.1292,
+      "step": 28700
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00011440677966101696,
+      "loss": 0.1492,
+      "step": 28710
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00011427009294696556,
+      "loss": 0.1123,
+      "step": 28720
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00011413340623291417,
+      "loss": 0.1325,
+      "step": 28730
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00011399671951886277,
+      "loss": 0.144,
+      "step": 28740
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00011386003280481137,
+      "loss": 0.1349,
+      "step": 28750
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00011372334609075997,
+      "loss": 0.1613,
+      "step": 28760
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0001135866593767086,
+      "loss": 0.1529,
+      "step": 28770
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0001134499726626572,
+      "loss": 0.1456,
+      "step": 28780
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0001133132859486058,
+      "loss": 0.1275,
+      "step": 28790
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00011317659923455441,
+      "loss": 0.1008,
+      "step": 28800
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00011303991252050301,
+      "loss": 0.1274,
+      "step": 28810
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00011290322580645161,
+      "loss": 0.1698,
+      "step": 28820
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00011276653909240023,
+      "loss": 0.1454,
+      "step": 28830
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00011262985237834883,
+      "loss": 0.1333,
+      "step": 28840
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00011249316566429743,
+      "loss": 0.132,
+      "step": 28850
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00011235647895024603,
+      "loss": 0.1474,
+      "step": 28860
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00011221979223619465,
+      "loss": 0.1259,
+      "step": 28870
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00011208310552214325,
+      "loss": 0.1991,
+      "step": 28880
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00011194641880809185,
+      "loss": 0.1252,
+      "step": 28890
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00011180973209404047,
+      "loss": 0.1239,
+      "step": 28900
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00011167304537998907,
+      "loss": 0.1463,
+      "step": 28910
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00011153635866593767,
+      "loss": 0.1244,
+      "step": 28920
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00011139967195188627,
+      "loss": 0.1698,
+      "step": 28930
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00011126298523783489,
+      "loss": 0.1582,
+      "step": 28940
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00011112629852378349,
+      "loss": 0.1274,
+      "step": 28950
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00011098961180973208,
+      "loss": 0.1193,
+      "step": 28960
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00011085292509568071,
+      "loss": 0.1238,
+      "step": 28970
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00011071623838162931,
+      "loss": 0.1747,
+      "step": 28980
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00011057955166757791,
+      "loss": 0.1759,
+      "step": 28990
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00011044286495352652,
+      "loss": 0.1424,
+      "step": 29000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00011030617823947512,
+      "loss": 0.1611,
+      "step": 29010
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00011016949152542372,
+      "loss": 0.1354,
+      "step": 29020
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00011003280481137234,
+      "loss": 0.1877,
+      "step": 29030
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00010989611809732095,
+      "loss": 0.1468,
+      "step": 29040
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00010975943138326955,
+      "loss": 0.1305,
+      "step": 29050
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00010962274466921815,
+      "loss": 0.1479,
+      "step": 29060
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00010948605795516676,
+      "loss": 0.1172,
+      "step": 29070
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00010934937124111536,
+      "loss": 0.1159,
+      "step": 29080
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00010921268452706396,
+      "loss": 0.1676,
+      "step": 29090
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00010907599781301259,
+      "loss": 0.0939,
+      "step": 29100
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00010893931109896119,
+      "loss": 0.2062,
+      "step": 29110
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00010880262438490979,
+      "loss": 0.1538,
+      "step": 29120
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.00010866593767085839,
+      "loss": 0.1347,
+      "step": 29130
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.000108529250956807,
+      "loss": 0.1684,
+      "step": 29140
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0001083925642427556,
+      "loss": 0.1247,
+      "step": 29150
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0001082558775287042,
+      "loss": 0.1346,
+      "step": 29160
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00010811919081465283,
+      "loss": 0.1406,
+      "step": 29170
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00010798250410060143,
+      "loss": 0.1343,
+      "step": 29180
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00010784581738655003,
+      "loss": 0.1723,
+      "step": 29190
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00010770913067249863,
+      "loss": 0.1328,
+      "step": 29200
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00010757244395844724,
+      "loss": 0.1208,
+      "step": 29210
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00010743575724439584,
+      "loss": 0.1528,
+      "step": 29220
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00010729907053034445,
+      "loss": 0.1342,
+      "step": 29230
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00010716238381629307,
+      "loss": 0.1892,
+      "step": 29240
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00010702569710224167,
+      "loss": 0.1433,
+      "step": 29250
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00010688901038819027,
+      "loss": 0.1819,
+      "step": 29260
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00010675232367413888,
+      "loss": 0.1392,
+      "step": 29270
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00010661563696008748,
+      "loss": 0.1735,
+      "step": 29280
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00010647895024603608,
+      "loss": 0.1281,
+      "step": 29290
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00010634226353198469,
+      "loss": 0.1557,
+      "step": 29300
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0001062055768179333,
+      "loss": 0.1772,
+      "step": 29310
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0001060688901038819,
+      "loss": 0.1391,
+      "step": 29320
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0001059322033898305,
+      "loss": 0.1631,
+      "step": 29330
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00010579551667577912,
+      "loss": 0.1528,
+      "step": 29340
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00010565882996172772,
+      "loss": 0.1182,
+      "step": 29350
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00010552214324767632,
+      "loss": 0.1521,
+      "step": 29360
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00010538545653362494,
+      "loss": 0.1477,
+      "step": 29370
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00010524876981957354,
+      "loss": 0.1546,
+      "step": 29380
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00010511208310552214,
+      "loss": 0.1608,
+      "step": 29390
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00010497539639147074,
+      "loss": 0.1441,
+      "step": 29400
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00010483870967741936,
+      "loss": 0.1705,
+      "step": 29410
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00010470202296336796,
+      "loss": 0.1404,
+      "step": 29420
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010456533624931657,
+      "loss": 0.1613,
+      "step": 29430
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010442864953526518,
+      "loss": 0.145,
+      "step": 29440
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010429196282121378,
+      "loss": 0.1257,
+      "step": 29450
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010415527610716238,
+      "loss": 0.1259,
+      "step": 29460
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.00010401858939311098,
+      "loss": 0.1744,
+      "step": 29470
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0001038819026790596,
+      "loss": 0.1279,
+      "step": 29480
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0001037452159650082,
+      "loss": 0.144,
+      "step": 29490
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00010360852925095681,
+      "loss": 0.1284,
+      "step": 29500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00010347184253690542,
+      "loss": 0.1646,
+      "step": 29510
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00010333515582285402,
+      "loss": 0.1425,
+      "step": 29520
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00010319846910880262,
+      "loss": 0.1211,
+      "step": 29530
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00010306178239475124,
+      "loss": 0.1317,
+      "step": 29540
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00010292509568069984,
+      "loss": 0.1195,
+      "step": 29550
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00010278840896664845,
+      "loss": 0.1442,
+      "step": 29560
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00010265172225259705,
+      "loss": 0.1594,
+      "step": 29570
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00010251503553854566,
+      "loss": 0.13,
+      "step": 29580
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00010237834882449426,
+      "loss": 0.1619,
+      "step": 29590
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00010224166211044286,
+      "loss": 0.1589,
+      "step": 29600
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.00010210497539639147,
+      "loss": 0.1928,
+      "step": 29610
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.00010196828868234007,
+      "loss": 0.1775,
+      "step": 29620
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.00010183160196828869,
+      "loss": 0.1392,
+      "step": 29630
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0001016949152542373,
+      "loss": 0.157,
+      "step": 29640
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0001015582285401859,
+      "loss": 0.1273,
+      "step": 29650
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0001014215418261345,
+      "loss": 0.109,
+      "step": 29660
+    },
+    {
+      "epoch": 8.0,
+      "eval_Bleu_1": 0.0050588086505624725,
+      "eval_Bleu_2": 2.088130782775145e-11,
+      "eval_Bleu_3": 3.715336010341995e-14,
+      "eval_Bleu_4": 1.6715032690467638e-15,
+      "eval_ROUGE_L": 0.007610609244860661,
+      "eval_cer": 0.5769631468769778,
+      "eval_em": 0.46462488129154794,
+      "eval_f1": 0.6234348759376488,
+      "eval_loss": 0.9071521162986755,
+      "eval_runtime": 318.5231,
+      "eval_samples_per_second": 13.224,
+      "eval_steps_per_second": 2.647,
+      "eval_wer": 0.6777239709443099,
+      "step": 29664
+    }
+  ],
+  "max_steps": 37080,
+  "num_train_epochs": 10,
+  "total_flos": 7.716068177739725e+17,
+  "trial_name": null,
+  "trial_params": null
+}