diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10500 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "global_step": 17365,
+  "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": 10.6173,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 10.5994,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 10.5936,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 10.5525,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 10.5214,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 10.3845,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 9.9854,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 9.1209,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 8.4798,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 7.8074,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011,
+      "loss": 7.5012,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012,
+      "loss": 7.3823,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 7.0647,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 7.2592,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015,
+      "loss": 7.1762,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016,
+      "loss": 7.0139,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017,
+      "loss": 6.9767,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 7.0242,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019,
+      "loss": 6.9285,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 6.9331,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021,
+      "loss": 6.8345,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022,
+      "loss": 6.6742,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023,
+      "loss": 6.8299,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00024,
+      "loss": 6.7196,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025,
+      "loss": 6.5603,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 6.5966,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027,
+      "loss": 6.6431,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 6.3873,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029,
+      "loss": 6.3411,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003,
+      "loss": 6.1473,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031,
+      "loss": 6.2374,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00032,
+      "loss": 6.2448,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033,
+      "loss": 6.1181,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00034,
+      "loss": 6.0314,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035,
+      "loss": 5.8719,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 5.8483,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00037,
+      "loss": 5.8423,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038,
+      "loss": 5.8773,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 5.8504,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004,
+      "loss": 5.4737,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00041,
+      "loss": 5.6913,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00042,
+      "loss": 5.7444,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00043,
+      "loss": 5.6371,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00044,
+      "loss": 5.6059,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 5.3563,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00046,
+      "loss": 5.2794,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00047,
+      "loss": 5.3608,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048,
+      "loss": 5.5027,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00049,
+      "loss": 5.3745,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 5.313,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004998834770449779,
+      "loss": 5.1712,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004997669540899557,
+      "loss": 5.219,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004996504311349336,
+      "loss": 5.3264,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004995339081799114,
+      "loss": 5.0162,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004994173852248894,
+      "loss": 4.986,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004993008622698671,
+      "loss": 4.8857,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004991843393148451,
+      "loss": 4.8639,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004990678163598229,
+      "loss": 5.2248,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004989512934048008,
+      "loss": 4.8982,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004988347704497786,
+      "loss": 5.0063,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004987182474947565,
+      "loss": 4.7881,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004986017245397343,
+      "loss": 4.824,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004984852015847123,
+      "loss": 4.5973,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00049836867862969,
+      "loss": 4.5633,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000498252155674668,
+      "loss": 4.7742,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004981356327196458,
+      "loss": 4.5603,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004980191097646236,
+      "loss": 4.5855,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004979025868096015,
+      "loss": 4.5635,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004977860638545793,
+      "loss": 4.4144,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004976695408995572,
+      "loss": 4.6402,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000497553017944535,
+      "loss": 4.4075,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004974364949895129,
+      "loss": 4.7223,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004973199720344908,
+      "loss": 4.383,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004972034490794687,
+      "loss": 4.392,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004970869261244465,
+      "loss": 4.1944,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004969704031694244,
+      "loss": 4.3497,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004968538802144022,
+      "loss": 4.0244,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004967373572593801,
+      "loss": 4.2946,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004966208343043579,
+      "loss": 4.0822,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004965043113493358,
+      "loss": 4.0192,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004963877883943137,
+      "loss": 4.1678,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004962712654392916,
+      "loss": 3.9266,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004961547424842694,
+      "loss": 3.9863,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004960382195292473,
+      "loss": 3.769,
+      "step": 840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004959216965742251,
+      "loss": 3.8906,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000495805173619203,
+      "loss": 3.9499,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004956886506641808,
+      "loss": 3.8681,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004955721277091587,
+      "loss": 3.8131,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004954556047541366,
+      "loss": 4.111,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004953390817991144,
+      "loss": 3.5835,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004952225588440923,
+      "loss": 3.8712,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004951060358890702,
+      "loss": 3.5934,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000494989512934048,
+      "loss": 3.6938,
+      "step": 930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004948729899790259,
+      "loss": 3.7079,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004947564670240037,
+      "loss": 3.6767,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004946399440689816,
+      "loss": 3.731,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004945234211139595,
+      "loss": 3.5824,
+      "step": 970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004944068981589373,
+      "loss": 3.3612,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004942903752039151,
+      "loss": 3.3219,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004941738522488931,
+      "loss": 3.5787,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004940573292938709,
+      "loss": 3.4477,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004939408063388488,
+      "loss": 3.4272,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004938242833838266,
+      "loss": 3.5109,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004937077604288045,
+      "loss": 3.0641,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004935912374737823,
+      "loss": 3.3225,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004934747145187602,
+      "loss": 3.2164,
+      "step": 1060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000493358191563738,
+      "loss": 3.2713,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000493241668608716,
+      "loss": 3.2336,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004931251456536937,
+      "loss": 3.2406,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004930086226986717,
+      "loss": 3.2814,
+      "step": 1100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004928920997436495,
+      "loss": 3.237,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004927755767886274,
+      "loss": 3.1105,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004926590538336052,
+      "loss": 3.121,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004925425308785831,
+      "loss": 2.8725,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004924260079235609,
+      "loss": 3.2262,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004923094849685389,
+      "loss": 3.0991,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004921929620135166,
+      "loss": 2.8996,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004920764390584946,
+      "loss": 3.2996,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004919599161034724,
+      "loss": 3.0485,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004918433931484503,
+      "loss": 2.9699,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004917268701934281,
+      "loss": 2.8221,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000491610347238406,
+      "loss": 2.9691,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004914938242833838,
+      "loss": 2.8891,
+      "step": 1230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004913773013283617,
+      "loss": 2.8668,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004912607783733395,
+      "loss": 2.8655,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004911442554183175,
+      "loss": 2.8323,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004910277324632953,
+      "loss": 2.775,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004909112095082732,
+      "loss": 2.7235,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000490794686553251,
+      "loss": 2.7463,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004906781635982289,
+      "loss": 2.8081,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004905616406432067,
+      "loss": 2.7157,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004904451176881846,
+      "loss": 2.7714,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004903285947331624,
+      "loss": 2.9056,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004902120717781404,
+      "loss": 2.7927,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004900955488231182,
+      "loss": 2.718,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004899790258680961,
+      "loss": 2.7283,
+      "step": 1360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004898625029130739,
+      "loss": 2.6543,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004897459799580518,
+      "loss": 2.5825,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004896294570030296,
+      "loss": 2.6792,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004895129340480074,
+      "loss": 2.4749,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004893964110929853,
+      "loss": 2.7346,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004892798881379633,
+      "loss": 2.4702,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000489163365182941,
+      "loss": 2.4849,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000489046842227919,
+      "loss": 2.5751,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004889303192728968,
+      "loss": 2.7505,
+      "step": 1450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004888137963178746,
+      "loss": 2.6167,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004886972733628525,
+      "loss": 2.5749,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004885807504078303,
+      "loss": 2.4734,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004884642274528082,
+      "loss": 2.5399,
+      "step": 1490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000488347704497786,
+      "loss": 2.7156,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004882311815427639,
+      "loss": 2.6661,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004881146585877418,
+      "loss": 2.4842,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00048799813563271964,
+      "loss": 2.5076,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00048788161267769754,
+      "loss": 2.4286,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00048776508972267534,
+      "loss": 2.3796,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00048764856676765324,
+      "loss": 2.249,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004875320438126311,
+      "loss": 2.4257,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000487415520857609,
+      "loss": 2.4112,
+      "step": 1580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004872989979025868,
+      "loss": 2.2194,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004871824749475647,
+      "loss": 2.3593,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00048706595199254254,
+      "loss": 2.3137,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004869494290375204,
+      "loss": 2.3152,
+      "step": 1620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048683290608249823,
+      "loss": 2.293,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048671638312747614,
+      "loss": 2.3859,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000486599860172454,
+      "loss": 2.5195,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00048648333721743183,
+      "loss": 2.2668,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004863668142624097,
+      "loss": 2.3753,
+      "step": 1670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004862502913073876,
+      "loss": 2.4052,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048613376835236543,
+      "loss": 2.2952,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004860172453973433,
+      "loss": 2.2161,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00048590072244232113,
+      "loss": 2.4161,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00048578419948729903,
+      "loss": 2.2893,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004856676765322769,
+      "loss": 2.2338,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00048555115357725473,
+      "loss": 1.9338,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004854346306222326,
+      "loss": 2.1428,
+      "step": 1750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004853181076672105,
+      "loss": 2.0794,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004852015847121883,
+      "loss": 2.1585,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004850850617571662,
+      "loss": 2.0462,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000484968538802144,
+      "loss": 2.1014,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004848520158471219,
+      "loss": 2.268,
+      "step": 1800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004847354928920997,
+      "loss": 2.11,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004846189699370776,
+      "loss": 2.275,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00048450244698205547,
+      "loss": 1.9513,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004843859240270334,
+      "loss": 2.1157,
+      "step": 1840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00048426940107201117,
+      "loss": 2.243,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00048415287811698907,
+      "loss": 2.0249,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004840363551619669,
+      "loss": 2.2214,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004839198322069448,
+      "loss": 2.1383,
+      "step": 1880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004838033092519226,
+      "loss": 2.0717,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004836867862969005,
+      "loss": 2.0318,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048357026334187837,
+      "loss": 2.2616,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048345374038685627,
+      "loss": 2.2362,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00048333721743183406,
+      "loss": 1.9737,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00048322069447681197,
+      "loss": 1.8598,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004831041715217898,
+      "loss": 2.0784,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00048298764856676766,
+      "loss": 2.1202,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004828711256117455,
+      "loss": 2.1851,
+      "step": 1970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048275460265672336,
+      "loss": 1.999,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048263807970170126,
+      "loss": 2.1535,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048252155674667906,
+      "loss": 1.927,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00048240503379165696,
+      "loss": 1.8046,
+      "step": 2010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004822885108366348,
+      "loss": 2.2255,
+      "step": 2020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004821719878816127,
+      "loss": 1.8354,
+      "step": 2030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004820554649265905,
+      "loss": 1.9391,
+      "step": 2040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004819389419715684,
+      "loss": 2.16,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00048182241901654626,
+      "loss": 2.0378,
+      "step": 2060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00048170589606152416,
+      "loss": 1.7791,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00048158937310650195,
+      "loss": 1.9443,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00048147285015147985,
+      "loss": 2.1102,
+      "step": 2090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004813563271964577,
+      "loss": 1.9356,
+      "step": 2100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004812398042414356,
+      "loss": 2.0797,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004811232812864134,
+      "loss": 1.9049,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004810067583313913,
+      "loss": 2.0014,
+      "step": 2130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00048089023537636915,
+      "loss": 1.8359,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00048077371242134705,
+      "loss": 1.8638,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00048065718946632485,
+      "loss": 1.851,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00048054066651130275,
+      "loss": 1.813,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004804241435562806,
+      "loss": 1.8793,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00048030762060125845,
+      "loss": 1.8115,
+      "step": 2190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004801910976462363,
+      "loss": 1.89,
+      "step": 2200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004800745746912142,
+      "loss": 1.6868,
+      "step": 2210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00047995805173619205,
+      "loss": 1.9705,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004798415287811699,
+      "loss": 1.7701,
+      "step": 2230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00047972500582614774,
+      "loss": 1.8274,
+      "step": 2240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00047960848287112565,
+      "loss": 1.9784,
+      "step": 2250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004794919599161035,
+      "loss": 1.8581,
+      "step": 2260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00047937543696108134,
+      "loss": 1.7436,
+      "step": 2270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004792589140060592,
+      "loss": 1.8933,
+      "step": 2280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004791423910510371,
+      "loss": 1.8863,
+      "step": 2290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047902586809601494,
+      "loss": 1.6754,
+      "step": 2300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004789093451409928,
+      "loss": 1.8452,
+      "step": 2310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047879282218597064,
+      "loss": 1.7726,
+      "step": 2320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00047867629923094854,
+      "loss": 1.6902,
+      "step": 2330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004785597762759264,
+      "loss": 1.7203,
+      "step": 2340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00047844325332090424,
+      "loss": 1.9117,
+      "step": 2350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004783267303658821,
+      "loss": 1.7419,
+      "step": 2360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00047821020741086,
+      "loss": 1.6398,
+      "step": 2370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004780936844558378,
+      "loss": 1.8232,
+      "step": 2380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004779771615008157,
+      "loss": 1.811,
+      "step": 2390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00047786063854579353,
+      "loss": 1.7618,
+      "step": 2400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00047774411559077144,
+      "loss": 1.8015,
+      "step": 2410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00047762759263574923,
+      "loss": 1.6506,
+      "step": 2420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004775110696807271,
+      "loss": 1.584,
+      "step": 2430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000477394546725705,
+      "loss": 1.7166,
+      "step": 2440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00047727802377068283,
+      "loss": 1.6907,
+      "step": 2450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004771615008156607,
+      "loss": 1.6075,
+      "step": 2460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004770449778606385,
+      "loss": 1.6789,
+      "step": 2470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00047692845490561643,
+      "loss": 1.6016,
+      "step": 2480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004768119319505943,
+      "loss": 1.737,
+      "step": 2490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004766954089955721,
+      "loss": 1.8943,
+      "step": 2500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00047657888604055,
+      "loss": 1.8478,
+      "step": 2510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004764623630855279,
+      "loss": 1.6695,
+      "step": 2520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00047634584013050567,
+      "loss": 1.6879,
+      "step": 2530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0004762293171754836,
+      "loss": 1.6278,
+      "step": 2540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0004761127942204614,
+      "loss": 1.7355,
+      "step": 2550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0004759962712654393,
+      "loss": 1.8658,
+      "step": 2560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0004758797483104171,
+      "loss": 1.6834,
+      "step": 2570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000475763225355395,
+      "loss": 1.6484,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00047564670240037287,
+      "loss": 1.6275,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00047553017944535077,
+      "loss": 1.6433,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00047541365649032857,
+      "loss": 1.6679,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00047529713353530647,
+      "loss": 1.615,
+      "step": 2620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0004751806105802843,
+      "loss": 1.6623,
+      "step": 2630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0004750640876252622,
+      "loss": 1.5745,
+      "step": 2640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00047494756467024,
+      "loss": 1.3781,
+      "step": 2650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0004748310417152179,
+      "loss": 1.6752,
+      "step": 2660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00047471451876019577,
+      "loss": 1.7025,
+      "step": 2670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00047459799580517367,
+      "loss": 1.6855,
+      "step": 2680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00047448147285015146,
+      "loss": 1.6679,
+      "step": 2690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00047436494989512936,
+      "loss": 1.6746,
+      "step": 2700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0004742484269401072,
+      "loss": 1.5759,
+      "step": 2710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00047413190398508506,
+      "loss": 1.3709,
+      "step": 2720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0004740153810300629,
+      "loss": 1.7639,
+      "step": 2730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0004738988580750408,
+      "loss": 1.3738,
+      "step": 2740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00047378233512001866,
+      "loss": 1.4433,
+      "step": 2750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0004736658121649965,
+      "loss": 1.4435,
+      "step": 2760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00047354928920997436,
+      "loss": 1.4654,
+      "step": 2770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00047343276625495226,
+      "loss": 1.5323,
+      "step": 2780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0004733162432999301,
+      "loss": 1.7076,
+      "step": 2790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00047319972034490796,
+      "loss": 1.6969,
+      "step": 2800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004730831973898858,
+      "loss": 1.7028,
+      "step": 2810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004729666744348637,
+      "loss": 1.5403,
+      "step": 2820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00047285015147984156,
+      "loss": 1.5397,
+      "step": 2830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004727336285248194,
+      "loss": 1.5315,
+      "step": 2840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00047261710556979725,
+      "loss": 1.4734,
+      "step": 2850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00047250058261477516,
+      "loss": 1.7396,
+      "step": 2860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000472384059659753,
+      "loss": 1.476,
+      "step": 2870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00047226753670473085,
+      "loss": 1.5518,
+      "step": 2880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004721510137497087,
+      "loss": 1.3857,
+      "step": 2890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004720344907946866,
+      "loss": 1.6634,
+      "step": 2900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004719179678396644,
+      "loss": 1.6851,
+      "step": 2910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00047180144488464225,
+      "loss": 1.5682,
+      "step": 2920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00047168492192962015,
+      "loss": 1.4285,
+      "step": 2930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000471568398974598,
+      "loss": 1.53,
+      "step": 2940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047145187601957585,
+      "loss": 1.6139,
+      "step": 2950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004713353530645537,
+      "loss": 1.7024,
+      "step": 2960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004712188301095316,
+      "loss": 1.5183,
+      "step": 2970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047110230715450944,
+      "loss": 1.6133,
+      "step": 2980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004709857841994873,
+      "loss": 1.7082,
+      "step": 2990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047086926124446514,
+      "loss": 1.4825,
+      "step": 3000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047075273828944304,
+      "loss": 1.3294,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004706362153344209,
+      "loss": 1.5075,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00047051969237939874,
+      "loss": 1.5766,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004704031694243766,
+      "loss": 1.5618,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004702866464693545,
+      "loss": 1.6336,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00047017012351433234,
+      "loss": 1.4952,
+      "step": 3060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004700536005593102,
+      "loss": 1.4659,
+      "step": 3070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00046993707760428804,
+      "loss": 1.6056,
+      "step": 3080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00046982055464926594,
+      "loss": 1.5604,
+      "step": 3090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00046970403169424373,
+      "loss": 1.5006,
+      "step": 3100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00046958750873922164,
+      "loss": 1.431,
+      "step": 3110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004694709857841995,
+      "loss": 1.3415,
+      "step": 3120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004693544628291774,
+      "loss": 1.5172,
+      "step": 3130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004692379398741552,
+      "loss": 1.3327,
+      "step": 3140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004691214169191331,
+      "loss": 1.4461,
+      "step": 3150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046900489396411093,
+      "loss": 1.3816,
+      "step": 3160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046888837100908884,
+      "loss": 1.5017,
+      "step": 3170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046877184805406663,
+      "loss": 1.3864,
+      "step": 3180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00046865532509904453,
+      "loss": 1.5376,
+      "step": 3190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004685388021440224,
+      "loss": 1.3746,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004684222791890003,
+      "loss": 1.2224,
+      "step": 3210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0004683057562339781,
+      "loss": 1.3624,
+      "step": 3220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000468189233278956,
+      "loss": 1.5236,
+      "step": 3230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00046807271032393383,
+      "loss": 1.3457,
+      "step": 3240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00046795618736891173,
+      "loss": 1.5382,
+      "step": 3250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0004678396644138895,
+      "loss": 1.4564,
+      "step": 3260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00046772314145886743,
+      "loss": 1.3994,
+      "step": 3270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004676066185038453,
+      "loss": 1.3252,
+      "step": 3280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004674900955488231,
+      "loss": 1.464,
+      "step": 3290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00046737357259380097,
+      "loss": 1.2866,
+      "step": 3300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004672570496387789,
+      "loss": 1.566,
+      "step": 3310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004671405266837567,
+      "loss": 1.5683,
+      "step": 3320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00046702400372873457,
+      "loss": 1.2691,
+      "step": 3330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004669074807737124,
+      "loss": 1.3731,
+      "step": 3340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004667909578186903,
+      "loss": 1.3997,
+      "step": 3350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00046667443486366817,
+      "loss": 1.5178,
+      "step": 3360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000466557911908646,
+      "loss": 1.4758,
+      "step": 3370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00046644138895362387,
+      "loss": 1.2601,
+      "step": 3380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00046632486599860177,
+      "loss": 1.5102,
+      "step": 3390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0004662083430435796,
+      "loss": 1.5268,
+      "step": 3400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0004660918200885574,
+      "loss": 1.4082,
+      "step": 3410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0004659752971335353,
+      "loss": 1.3945,
+      "step": 3420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00046585877417851316,
+      "loss": 1.4644,
+      "step": 3430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00046574225122349107,
+      "loss": 1.3928,
+      "step": 3440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00046562572826846886,
+      "loss": 1.3034,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00046550920531344676,
+      "loss": 1.3036,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0004653926823584246,
+      "loss": 1.5186,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00046527615940340246,
+      "loss": 1.3525,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0004651596364483803,
+      "loss": 1.4504,
+      "step": 3490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0004650431134933582,
+      "loss": 1.2229,
+      "step": 3500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00046492659053833606,
+      "loss": 1.2556,
+      "step": 3510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0004648100675833139,
+      "loss": 1.2992,
+      "step": 3520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00046469354462829176,
+      "loss": 1.256,
+      "step": 3530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00046457702167326966,
+      "loss": 1.4052,
+      "step": 3540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0004644604987182475,
+      "loss": 1.4992,
+      "step": 3550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00046434397576322536,
+      "loss": 1.2673,
+      "step": 3560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0004642274528082032,
+      "loss": 1.3817,
+      "step": 3570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0004641109298531811,
+      "loss": 1.159,
+      "step": 3580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00046399440689815896,
+      "loss": 1.3467,
+      "step": 3590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0004638778839431368,
+      "loss": 1.3512,
+      "step": 3600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00046376136098811465,
+      "loss": 1.31,
+      "step": 3610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00046364483803309255,
+      "loss": 1.2474,
+      "step": 3620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0004635283150780704,
+      "loss": 1.3939,
+      "step": 3630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00046341179212304825,
+      "loss": 1.3823,
+      "step": 3640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0004632952691680261,
+      "loss": 1.3648,
+      "step": 3650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000463178746213004,
+      "loss": 1.1649,
+      "step": 3660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0004630622232579818,
+      "loss": 1.2892,
+      "step": 3670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0004629457003029597,
+      "loss": 1.3038,
+      "step": 3680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046282917734793755,
+      "loss": 1.2314,
+      "step": 3690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046271265439291545,
+      "loss": 1.1314,
+      "step": 3700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00046259613143789324,
+      "loss": 1.3848,
+      "step": 3710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00046247960848287115,
+      "loss": 1.3334,
+      "step": 3720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000462363085527849,
+      "loss": 1.2365,
+      "step": 3730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0004622465625728269,
+      "loss": 1.2673,
+      "step": 3740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0004621300396178047,
+      "loss": 1.2443,
+      "step": 3750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0004620135166627826,
+      "loss": 1.2275,
+      "step": 3760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00046189699370776044,
+      "loss": 1.3776,
+      "step": 3770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00046178047075273835,
+      "loss": 1.4364,
+      "step": 3780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00046166394779771614,
+      "loss": 1.3173,
+      "step": 3790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00046154742484269404,
+      "loss": 1.3461,
+      "step": 3800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004614309018876719,
+      "loss": 1.3649,
+      "step": 3810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00046131437893264974,
+      "loss": 1.3683,
+      "step": 3820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004611978559776276,
+      "loss": 1.3727,
+      "step": 3830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004610813330226055,
+      "loss": 1.1828,
+      "step": 3840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00046096481006758334,
+      "loss": 1.19,
+      "step": 3850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0004608482871125612,
+      "loss": 1.3646,
+      "step": 3860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00046073176415753904,
+      "loss": 1.1841,
+      "step": 3870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0004606152412025169,
+      "loss": 1.1462,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0004604987182474948,
+      "loss": 1.1347,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0004603821952924726,
+      "loss": 1.17,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0004602656723374505,
+      "loss": 1.2542,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00046014914938242833,
+      "loss": 1.4318,
+      "step": 3920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00046003262642740623,
+      "loss": 1.3082,
+      "step": 3930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00045991610347238403,
+      "loss": 1.2716,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00045979958051736193,
+      "loss": 1.2412,
+      "step": 3950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0004596830575623398,
+      "loss": 1.3664,
+      "step": 3960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0004595665346073177,
+      "loss": 1.2538,
+      "step": 3970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0004594500116522955,
+      "loss": 1.2536,
+      "step": 3980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0004593334886972734,
+      "loss": 1.1971,
+      "step": 3990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0004592169657422512,
+      "loss": 1.2953,
+      "step": 4000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0004591004427872291,
+      "loss": 1.158,
+      "step": 4010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004589839198322069,
+      "loss": 1.5582,
+      "step": 4020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004588673968771848,
+      "loss": 1.1798,
+      "step": 4030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004587508739221627,
+      "loss": 1.184,
+      "step": 4040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004586343509671405,
+      "loss": 1.3451,
+      "step": 4050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00045851782801211837,
+      "loss": 1.2911,
+      "step": 4060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0004584013050570963,
+      "loss": 1.3576,
+      "step": 4070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0004582847821020741,
+      "loss": 1.3262,
+      "step": 4080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00045816825914705197,
+      "loss": 1.2768,
+      "step": 4090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0004580517361920298,
+      "loss": 1.1113,
+      "step": 4100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0004579352132370077,
+      "loss": 1.3998,
+      "step": 4110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00045781869028198557,
+      "loss": 1.4512,
+      "step": 4120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0004577021673269634,
+      "loss": 1.2316,
+      "step": 4130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00045758564437194127,
+      "loss": 1.1764,
+      "step": 4140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00045746912141691917,
+      "loss": 1.0234,
+      "step": 4150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000457352598461897,
+      "loss": 1.2533,
+      "step": 4160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00045723607550687487,
+      "loss": 1.1746,
+      "step": 4170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0004571195525518527,
+      "loss": 1.057,
+      "step": 4180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004570030295968306,
+      "loss": 1.2712,
+      "step": 4190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004568865066418084,
+      "loss": 1.1473,
+      "step": 4200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004567699836867863,
+      "loss": 1.3898,
+      "step": 4210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00045665346073176416,
+      "loss": 1.3249,
+      "step": 4220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00045653693777674206,
+      "loss": 1.2951,
+      "step": 4230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00045642041482171986,
+      "loss": 1.1313,
+      "step": 4240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00045630389186669776,
+      "loss": 1.234,
+      "step": 4250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0004561873689116756,
+      "loss": 1.2596,
+      "step": 4260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0004560708459566535,
+      "loss": 1.2363,
+      "step": 4270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0004559543230016313,
+      "loss": 1.0825,
+      "step": 4280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0004558378000466092,
+      "loss": 1.1746,
+      "step": 4290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00045572127709158706,
+      "loss": 1.3537,
+      "step": 4300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00045560475413656496,
+      "loss": 1.1291,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00045548823118154275,
+      "loss": 1.165,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00045537170822652066,
+      "loss": 1.2263,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004552551852714985,
+      "loss": 1.1819,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0,
+      "eval_Bleu_1": 0.004562230879638888,
+      "eval_Bleu_2": 1.832697466618859e-11,
+      "eval_Bleu_3": 3.2617476367192263e-14,
+      "eval_Bleu_4": 1.4785080931925696e-15,
+      "eval_ROUGE_L": 0.007574926108851014,
+      "eval_cer": 0.6058397695082681,
+      "eval_em": 0.4438259109311741,
+      "eval_f1": 0.6230237395969866,
+      "eval_loss": 1.157799243927002,
+      "eval_runtime": 355.2702,
+      "eval_samples_per_second": 16.686,
+      "eval_steps_per_second": 3.338,
+      "eval_wer": 0.7307975977241598,
+      "step": 4341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004551386623164764,
+      "loss": 1.0483,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004550221393614542,
+      "loss": 0.9632,
+      "step": 4360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00045490561640643205,
+      "loss": 1.0083,
+      "step": 4370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00045478909345140995,
+      "loss": 0.861,
+      "step": 4380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00045467257049638775,
+      "loss": 0.9485,
+      "step": 4390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00045455604754136565,
+      "loss": 0.899,
+      "step": 4400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004544395245863435,
+      "loss": 0.9807,
+      "step": 4410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004543230016313214,
+      "loss": 0.9033,
+      "step": 4420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004542064786762992,
+      "loss": 0.8492,
+      "step": 4430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004540899557212771,
+      "loss": 0.9586,
+      "step": 4440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00045397343276625495,
+      "loss": 0.9832,
+      "step": 4450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00045385690981123285,
+      "loss": 0.8696,
+      "step": 4460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00045374038685621064,
+      "loss": 0.8504,
+      "step": 4470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00045362386390118855,
+      "loss": 0.8982,
+      "step": 4480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004535073409461664,
+      "loss": 0.9312,
+      "step": 4490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004533908179911443,
+      "loss": 0.8821,
+      "step": 4500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004532742950361221,
+      "loss": 0.9766,
+      "step": 4510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004531577720811,
+      "loss": 0.8721,
+      "step": 4520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00045304124912607784,
+      "loss": 0.9112,
+      "step": 4530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00045292472617105574,
+      "loss": 0.9845,
+      "step": 4540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00045280820321603354,
+      "loss": 0.8542,
+      "step": 4550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00045269168026101144,
+      "loss": 0.7962,
+      "step": 4560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004525751573059893,
+      "loss": 1.0538,
+      "step": 4570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00045245863435096714,
+      "loss": 0.8934,
+      "step": 4580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000452342111395945,
+      "loss": 0.7979,
+      "step": 4590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004522255884409229,
+      "loss": 0.8273,
+      "step": 4600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00045210906548590074,
+      "loss": 0.9187,
+      "step": 4610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004519925425308786,
+      "loss": 0.8956,
+      "step": 4620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00045187601957585643,
+      "loss": 0.8595,
+      "step": 4630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00045175949662083434,
+      "loss": 0.9169,
+      "step": 4640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004516429736658122,
+      "loss": 1.0031,
+      "step": 4650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00045152645071079003,
+      "loss": 0.9034,
+      "step": 4660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004514099277557679,
+      "loss": 0.9478,
+      "step": 4670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004512934048007458,
+      "loss": 1.0032,
+      "step": 4680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045117688184572363,
+      "loss": 0.8712,
+      "step": 4690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004510603588907015,
+      "loss": 1.0547,
+      "step": 4700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045094383593567933,
+      "loss": 0.8324,
+      "step": 4710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045082731298065723,
+      "loss": 0.8245,
+      "step": 4720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004507107900256351,
+      "loss": 0.9643,
+      "step": 4730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045059426707061293,
+      "loss": 0.9543,
+      "step": 4740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004504777441155908,
+      "loss": 0.8372,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004503612211605687,
+      "loss": 0.8607,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004502446982055465,
+      "loss": 0.9938,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004501281752505244,
+      "loss": 0.847,
+      "step": 4780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004500116522955022,
+      "loss": 1.0042,
+      "step": 4790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00044989512934048013,
+      "loss": 0.798,
+      "step": 4800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004497786063854579,
+      "loss": 1.0318,
+      "step": 4810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004496620834304358,
+      "loss": 0.9688,
+      "step": 4820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00044954556047541367,
+      "loss": 0.9583,
+      "step": 4830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004494290375203916,
+      "loss": 0.8226,
+      "step": 4840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00044931251456536937,
+      "loss": 0.9248,
+      "step": 4850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004491959916103472,
+      "loss": 0.8285,
+      "step": 4860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004490794686553251,
+      "loss": 0.7634,
+      "step": 4870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00044896294570030297,
+      "loss": 0.8457,
+      "step": 4880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004488464227452808,
+      "loss": 1.01,
+      "step": 4890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00044872989979025867,
+      "loss": 0.8146,
+      "step": 4900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00044861337683523657,
+      "loss": 0.895,
+      "step": 4910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00044849685388021436,
+      "loss": 0.8456,
+      "step": 4920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00044838033092519226,
+      "loss": 0.8982,
+      "step": 4930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004482638079701701,
+      "loss": 1.0002,
+      "step": 4940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000448147285015148,
+      "loss": 0.9766,
+      "step": 4950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004480307620601258,
+      "loss": 0.8912,
+      "step": 4960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004479142391051037,
+      "loss": 0.9753,
+      "step": 4970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00044779771615008156,
+      "loss": 0.9715,
+      "step": 4980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00044768119319505946,
+      "loss": 0.9285,
+      "step": 4990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00044756467024003726,
+      "loss": 0.9086,
+      "step": 5000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00044744814728501516,
+      "loss": 0.9184,
+      "step": 5010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000447331624329993,
+      "loss": 0.7753,
+      "step": 5020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004472151013749709,
+      "loss": 0.8555,
+      "step": 5030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004470985784199487,
+      "loss": 0.8246,
+      "step": 5040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0004469820554649266,
+      "loss": 0.9383,
+      "step": 5050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00044686553250990446,
+      "loss": 0.9073,
+      "step": 5060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00044674900955488236,
+      "loss": 0.8755,
+      "step": 5070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00044663248659986015,
+      "loss": 0.9361,
+      "step": 5080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00044651596364483806,
+      "loss": 0.8497,
+      "step": 5090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0004463994406898159,
+      "loss": 1.0263,
+      "step": 5100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00044628291773479375,
+      "loss": 1.0664,
+      "step": 5110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0004461663947797716,
+      "loss": 0.9687,
+      "step": 5120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0004460498718247495,
+      "loss": 0.9282,
+      "step": 5130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00044593334886972735,
+      "loss": 0.9985,
+      "step": 5140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0004458168259147052,
+      "loss": 0.9896,
+      "step": 5150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00044570030295968305,
+      "loss": 0.9017,
+      "step": 5160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00044558378000466095,
+      "loss": 0.8536,
+      "step": 5170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0004454672570496388,
+      "loss": 0.6861,
+      "step": 5180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00044535073409461665,
+      "loss": 0.9811,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0004452342111395945,
+      "loss": 0.8626,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0004451176881845724,
+      "loss": 0.8912,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00044500116522955025,
+      "loss": 0.9315,
+      "step": 5220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0004448846422745281,
+      "loss": 0.9026,
+      "step": 5230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00044476811931950594,
+      "loss": 0.9576,
+      "step": 5240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00044465159636448385,
+      "loss": 0.9743,
+      "step": 5250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0004445350734094617,
+      "loss": 0.8514,
+      "step": 5260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00044441855045443954,
+      "loss": 0.7854,
+      "step": 5270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004443020274994174,
+      "loss": 0.9221,
+      "step": 5280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004441855045443953,
+      "loss": 0.9368,
+      "step": 5290
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004440689815893731,
+      "loss": 0.8692,
+      "step": 5300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000443952458634351,
+      "loss": 0.94,
+      "step": 5310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00044383593567932884,
+      "loss": 0.888,
+      "step": 5320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0004437194127243067,
+      "loss": 0.8532,
+      "step": 5330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00044360288976928454,
+      "loss": 0.8541,
+      "step": 5340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0004434863668142624,
+      "loss": 0.7617,
+      "step": 5350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0004433698438592403,
+      "loss": 0.9921,
+      "step": 5360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00044325332090421814,
+      "loss": 0.9467,
+      "step": 5370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000443136797949196,
+      "loss": 1.0556,
+      "step": 5380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00044302027499417383,
+      "loss": 0.8402,
+      "step": 5390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00044290375203915173,
+      "loss": 0.9111,
+      "step": 5400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004427872290841296,
+      "loss": 0.9139,
+      "step": 5410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00044267070612910743,
+      "loss": 0.8888,
+      "step": 5420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004425541831740853,
+      "loss": 0.8642,
+      "step": 5430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004424376602190632,
+      "loss": 1.0252,
+      "step": 5440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00044232113726404103,
+      "loss": 0.8512,
+      "step": 5450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0004422046143090189,
+      "loss": 0.8371,
+      "step": 5460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00044208809135399673,
+      "loss": 0.9973,
+      "step": 5470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00044197156839897463,
+      "loss": 0.9172,
+      "step": 5480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0004418550454439524,
+      "loss": 0.9916,
+      "step": 5490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0004417385224889303,
+      "loss": 1.0238,
+      "step": 5500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0004416219995339082,
+      "loss": 0.8475,
+      "step": 5510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0004415054765788861,
+      "loss": 0.9013,
+      "step": 5520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00044138895362386387,
+      "loss": 0.9205,
+      "step": 5530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004412724306688418,
+      "loss": 0.7697,
+      "step": 5540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004411559077138196,
+      "loss": 0.9124,
+      "step": 5550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004410393847587975,
+      "loss": 1.0096,
+      "step": 5560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004409228618037753,
+      "loss": 1.0114,
+      "step": 5570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0004408063388487532,
+      "loss": 0.8859,
+      "step": 5580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00044068981589373107,
+      "loss": 1.1045,
+      "step": 5590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000440573292938709,
+      "loss": 0.8819,
+      "step": 5600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00044045676998368677,
+      "loss": 0.9543,
+      "step": 5610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00044034024702866467,
+      "loss": 0.8881,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004402237240736425,
+      "loss": 0.8167,
+      "step": 5630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004401072011186204,
+      "loss": 0.8464,
+      "step": 5640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004399906781635982,
+      "loss": 0.8874,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004398741552085761,
+      "loss": 0.9212,
+      "step": 5660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00043975763225355397,
+      "loss": 0.868,
+      "step": 5670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004396411092985318,
+      "loss": 0.9033,
+      "step": 5680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00043952458634350966,
+      "loss": 0.9567,
+      "step": 5690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00043940806338848757,
+      "loss": 0.9298,
+      "step": 5700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0004392915404334654,
+      "loss": 0.8746,
+      "step": 5710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00043917501747844326,
+      "loss": 0.9638,
+      "step": 5720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0004390584945234211,
+      "loss": 0.8256,
+      "step": 5730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000438941971568399,
+      "loss": 0.9559,
+      "step": 5740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00043882544861337686,
+      "loss": 0.866,
+      "step": 5750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004387089256583547,
+      "loss": 0.83,
+      "step": 5760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00043859240270333256,
+      "loss": 0.8454,
+      "step": 5770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00043847587974831046,
+      "loss": 0.8315,
+      "step": 5780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004383593567932883,
+      "loss": 0.9442,
+      "step": 5790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0004382428338382661,
+      "loss": 0.808,
+      "step": 5800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000438126310883244,
+      "loss": 1.0284,
+      "step": 5810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00043800978792822185,
+      "loss": 0.882,
+      "step": 5820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00043789326497319976,
+      "loss": 0.9005,
+      "step": 5830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00043777674201817755,
+      "loss": 0.8253,
+      "step": 5840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00043766021906315545,
+      "loss": 0.995,
+      "step": 5850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0004375436961081333,
+      "loss": 0.8564,
+      "step": 5860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00043742717315311115,
+      "loss": 1.0516,
+      "step": 5870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000437310650198089,
+      "loss": 0.8854,
+      "step": 5880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0004371941272430669,
+      "loss": 0.9493,
+      "step": 5890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00043707760428804475,
+      "loss": 0.9142,
+      "step": 5900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0004369610813330226,
+      "loss": 0.8572,
+      "step": 5910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00043684455837800045,
+      "loss": 0.7193,
+      "step": 5920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00043672803542297835,
+      "loss": 0.9045,
+      "step": 5930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0004366115124679562,
+      "loss": 0.9262,
+      "step": 5940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00043649498951293405,
+      "loss": 0.7985,
+      "step": 5950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0004363784665579119,
+      "loss": 0.7903,
+      "step": 5960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0004362619436028898,
+      "loss": 0.8705,
+      "step": 5970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00043614542064786765,
+      "loss": 0.9021,
+      "step": 5980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0004360288976928455,
+      "loss": 0.7522,
+      "step": 5990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00043591237473782334,
+      "loss": 0.9521,
+      "step": 6000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00043579585178280124,
+      "loss": 0.9453,
+      "step": 6010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0004356793288277791,
+      "loss": 0.81,
+      "step": 6020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00043556280587275694,
+      "loss": 0.9281,
+      "step": 6030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0004354462829177348,
+      "loss": 0.8549,
+      "step": 6040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0004353297599627127,
+      "loss": 0.9825,
+      "step": 6050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0004352132370076905,
+      "loss": 0.945,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0004350967140526684,
+      "loss": 0.6956,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00043498019109764624,
+      "loss": 0.9248,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00043486366814262414,
+      "loss": 0.8503,
+      "step": 6090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00043474714518760193,
+      "loss": 0.9704,
+      "step": 6100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00043463062223257984,
+      "loss": 0.9158,
+      "step": 6110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0004345140992775577,
+      "loss": 0.8862,
+      "step": 6120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0004343975763225356,
+      "loss": 0.7859,
+      "step": 6130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0004342810533675134,
+      "loss": 0.8887,
+      "step": 6140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0004341645304124913,
+      "loss": 0.7453,
+      "step": 6150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00043404800745746913,
+      "loss": 0.8689,
+      "step": 6160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00043393148450244704,
+      "loss": 0.7721,
+      "step": 6170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00043381496154742483,
+      "loss": 0.843,
+      "step": 6180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00043369843859240273,
+      "loss": 0.9345,
+      "step": 6190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004335819156373806,
+      "loss": 0.8555,
+      "step": 6200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00043346539268235843,
+      "loss": 0.8983,
+      "step": 6210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004333488697273363,
+      "loss": 0.8739,
+      "step": 6220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004332323467723142,
+      "loss": 0.9557,
+      "step": 6230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00043311582381729203,
+      "loss": 0.9242,
+      "step": 6240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004329993008622699,
+      "loss": 0.778,
+      "step": 6250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004328827779072477,
+      "loss": 0.8425,
+      "step": 6260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00043276625495222563,
+      "loss": 0.8424,
+      "step": 6270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0004326497319972035,
+      "loss": 0.9171,
+      "step": 6280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00043253320904218127,
+      "loss": 0.9715,
+      "step": 6290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0004324166860871592,
+      "loss": 0.7575,
+      "step": 6300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000432300163132137,
+      "loss": 0.8489,
+      "step": 6310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004321836401771149,
+      "loss": 0.8243,
+      "step": 6320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004320671172220927,
+      "loss": 0.8604,
+      "step": 6330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004319505942670706,
+      "loss": 0.9974,
+      "step": 6340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00043183407131204847,
+      "loss": 0.9065,
+      "step": 6350
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00043171754835702637,
+      "loss": 0.8044,
+      "step": 6360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00043160102540200417,
+      "loss": 0.9111,
+      "step": 6370
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00043148450244698207,
+      "loss": 0.9364,
+      "step": 6380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0004313679794919599,
+      "loss": 0.8497,
+      "step": 6390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00043125145653693777,
+      "loss": 0.8877,
+      "step": 6400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004311349335819156,
+      "loss": 0.8173,
+      "step": 6410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004310184106268935,
+      "loss": 0.946,
+      "step": 6420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00043090188767187136,
+      "loss": 0.9376,
+      "step": 6430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004307853647168492,
+      "loss": 0.85,
+      "step": 6440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00043066884176182706,
+      "loss": 0.8629,
+      "step": 6450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00043055231880680496,
+      "loss": 0.8033,
+      "step": 6460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004304357958517828,
+      "loss": 1.0201,
+      "step": 6470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00043031927289676066,
+      "loss": 0.8825,
+      "step": 6480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004302027499417385,
+      "loss": 0.8825,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0004300862269867164,
+      "loss": 0.7078,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00042996970403169426,
+      "loss": 0.8709,
+      "step": 6510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0004298531810766721,
+      "loss": 0.8811,
+      "step": 6520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00042973665812164996,
+      "loss": 0.7586,
+      "step": 6530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00042962013516662786,
+      "loss": 0.925,
+      "step": 6540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0004295036122116057,
+      "loss": 0.8418,
+      "step": 6550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00042938708925658356,
+      "loss": 0.806,
+      "step": 6560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0004292705663015614,
+      "loss": 0.864,
+      "step": 6570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0004291540433465393,
+      "loss": 0.8085,
+      "step": 6580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0004290375203915171,
+      "loss": 1.0106,
+      "step": 6590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000428920997436495,
+      "loss": 0.9069,
+      "step": 6600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00042880447448147285,
+      "loss": 0.8129,
+      "step": 6610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00042868795152645076,
+      "loss": 0.846,
+      "step": 6620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00042857142857142855,
+      "loss": 0.8213,
+      "step": 6630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00042845490561640645,
+      "loss": 0.821,
+      "step": 6640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0004283383826613843,
+      "loss": 0.8194,
+      "step": 6650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0004282218597063622,
+      "loss": 0.7394,
+      "step": 6660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00042810533675134,
+      "loss": 0.9139,
+      "step": 6670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0004279888137963179,
+      "loss": 0.8334,
+      "step": 6680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00042787229084129575,
+      "loss": 0.7381,
+      "step": 6690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00042775576788627365,
+      "loss": 0.7279,
+      "step": 6700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00042763924493125144,
+      "loss": 0.9201,
+      "step": 6710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00042752272197622935,
+      "loss": 0.8286,
+      "step": 6720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0004274061990212072,
+      "loss": 0.8908,
+      "step": 6730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0004272896760661851,
+      "loss": 0.8951,
+      "step": 6740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0004271731531111629,
+      "loss": 0.8557,
+      "step": 6750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0004270566301561408,
+      "loss": 0.8737,
+      "step": 6760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00042694010720111864,
+      "loss": 0.8117,
+      "step": 6770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00042682358424609644,
+      "loss": 0.8245,
+      "step": 6780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00042670706129107434,
+      "loss": 0.8187,
+      "step": 6790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004265905383360522,
+      "loss": 0.7613,
+      "step": 6800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004264740153810301,
+      "loss": 0.9373,
+      "step": 6810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004263574924260079,
+      "loss": 0.9169,
+      "step": 6820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004262409694709858,
+      "loss": 0.831,
+      "step": 6830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00042612444651596364,
+      "loss": 0.8466,
+      "step": 6840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00042600792356094154,
+      "loss": 0.8031,
+      "step": 6850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00042589140060591933,
+      "loss": 1.0256,
+      "step": 6860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00042577487765089724,
+      "loss": 0.8324,
+      "step": 6870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0004256583546958751,
+      "loss": 0.8445,
+      "step": 6880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000425541831740853,
+      "loss": 0.8516,
+      "step": 6890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0004254253087858308,
+      "loss": 0.6436,
+      "step": 6900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0004253087858308087,
+      "loss": 0.8925,
+      "step": 6910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00042519226287578653,
+      "loss": 0.8912,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00042507573992076443,
+      "loss": 0.959,
+      "step": 6930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00042495921696574223,
+      "loss": 0.9026,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00042484269401072013,
+      "loss": 0.8484,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000424726171055698,
+      "loss": 0.9602,
+      "step": 6960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00042460964810067583,
+      "loss": 0.7942,
+      "step": 6970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0004244931251456537,
+      "loss": 0.9194,
+      "step": 6980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0004243766021906316,
+      "loss": 0.8724,
+      "step": 6990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00042426007923560943,
+      "loss": 0.8077,
+      "step": 7000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0004241435562805873,
+      "loss": 0.8402,
+      "step": 7010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004240270333255651,
+      "loss": 0.8272,
+      "step": 7020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000423910510370543,
+      "loss": 0.8768,
+      "step": 7030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004237939874155209,
+      "loss": 0.8291,
+      "step": 7040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0004236774644604987,
+      "loss": 0.8288,
+      "step": 7050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00042356094150547657,
+      "loss": 0.8869,
+      "step": 7060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0004234444185504545,
+      "loss": 0.9815,
+      "step": 7070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0004233278955954323,
+      "loss": 0.8464,
+      "step": 7080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00042321137264041017,
+      "loss": 0.8099,
+      "step": 7090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000423094849685388,
+      "loss": 0.8248,
+      "step": 7100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004229783267303659,
+      "loss": 0.8718,
+      "step": 7110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00042286180377534377,
+      "loss": 0.674,
+      "step": 7120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004227452808203216,
+      "loss": 0.8456,
+      "step": 7130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00042262875786529947,
+      "loss": 0.8607,
+      "step": 7140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00042251223491027737,
+      "loss": 0.7995,
+      "step": 7150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00042239571195525516,
+      "loss": 0.8364,
+      "step": 7160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00042227918900023307,
+      "loss": 0.8575,
+      "step": 7170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004221626660452109,
+      "loss": 0.7937,
+      "step": 7180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004220461430901888,
+      "loss": 0.7883,
+      "step": 7190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004219296201351666,
+      "loss": 0.8459,
+      "step": 7200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004218130971801445,
+      "loss": 0.7338,
+      "step": 7210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00042169657422512236,
+      "loss": 0.8326,
+      "step": 7220
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00042158005127010027,
+      "loss": 0.7968,
+      "step": 7230
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00042146352831507806,
+      "loss": 0.8392,
+      "step": 7240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004213470053600559,
+      "loss": 0.7962,
+      "step": 7250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004212304824050338,
+      "loss": 0.8887,
+      "step": 7260
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00042111395945001166,
+      "loss": 0.8364,
+      "step": 7270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004209974364949895,
+      "loss": 0.8338,
+      "step": 7280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00042088091353996736,
+      "loss": 0.825,
+      "step": 7290
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00042076439058494526,
+      "loss": 0.7979,
+      "step": 7300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004206478676299231,
+      "loss": 0.8145,
+      "step": 7310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00042053134467490096,
+      "loss": 0.9179,
+      "step": 7320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004204148217198788,
+      "loss": 0.9284,
+      "step": 7330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004202982987648567,
+      "loss": 0.7933,
+      "step": 7340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004201817758098345,
+      "loss": 0.8116,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004200652528548124,
+      "loss": 0.9992,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00041994872989979025,
+      "loss": 0.7485,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00041983220694476815,
+      "loss": 0.8439,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00041971568398974595,
+      "loss": 0.8971,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00041959916103472385,
+      "loss": 0.8182,
+      "step": 7400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004194826380797017,
+      "loss": 0.9844,
+      "step": 7410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004193661151246796,
+      "loss": 0.7211,
+      "step": 7420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004192495921696574,
+      "loss": 0.9946,
+      "step": 7430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004191330692146353,
+      "loss": 0.7788,
+      "step": 7440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00041901654625961315,
+      "loss": 0.7851,
+      "step": 7450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00041890002330459105,
+      "loss": 0.8839,
+      "step": 7460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00041878350034956884,
+      "loss": 0.7923,
+      "step": 7470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00041866697739454675,
+      "loss": 0.9646,
+      "step": 7480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004185504544395246,
+      "loss": 0.8403,
+      "step": 7490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00041843393148450244,
+      "loss": 0.8778,
+      "step": 7500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004183174085294803,
+      "loss": 0.8131,
+      "step": 7510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004182008855744582,
+      "loss": 0.8474,
+      "step": 7520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00041808436261943604,
+      "loss": 0.742,
+      "step": 7530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004179678396644139,
+      "loss": 0.8989,
+      "step": 7540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00041785131670939174,
+      "loss": 0.9878,
+      "step": 7550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00041773479375436964,
+      "loss": 0.8157,
+      "step": 7560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004176182707993475,
+      "loss": 0.7535,
+      "step": 7570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00041750174784432534,
+      "loss": 0.7385,
+      "step": 7580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004173852248893032,
+      "loss": 0.7585,
+      "step": 7590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004172687019342811,
+      "loss": 0.7945,
+      "step": 7600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00041715217897925894,
+      "loss": 0.9057,
+      "step": 7610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004170356560242368,
+      "loss": 0.9158,
+      "step": 7620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00041691913306921463,
+      "loss": 0.7643,
+      "step": 7630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00041680261011419254,
+      "loss": 0.9082,
+      "step": 7640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004166860871591704,
+      "loss": 0.8074,
+      "step": 7650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00041656956420414823,
+      "loss": 0.8786,
+      "step": 7660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004164530412491261,
+      "loss": 0.8093,
+      "step": 7670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000416336518294104,
+      "loss": 0.8411,
+      "step": 7680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004162199953390818,
+      "loss": 0.8258,
+      "step": 7690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004161034723840597,
+      "loss": 0.8163,
+      "step": 7700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00041598694942903753,
+      "loss": 0.9144,
+      "step": 7710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00041587042647401543,
+      "loss": 0.7808,
+      "step": 7720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004157539035189932,
+      "loss": 0.7541,
+      "step": 7730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004156373805639711,
+      "loss": 0.7985,
+      "step": 7740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000415520857608949,
+      "loss": 0.7638,
+      "step": 7750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004154043346539268,
+      "loss": 0.7434,
+      "step": 7760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004152878116989047,
+      "loss": 0.7973,
+      "step": 7770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004151712887438825,
+      "loss": 0.7924,
+      "step": 7780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004150547657888604,
+      "loss": 0.8787,
+      "step": 7790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004149382428338383,
+      "loss": 0.8441,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004148217198788161,
+      "loss": 0.8471,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00041470519692379397,
+      "loss": 0.8283,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004145886739687719,
+      "loss": 0.7841,
+      "step": 7830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004144721510137497,
+      "loss": 0.8118,
+      "step": 7840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00041435562805872757,
+      "loss": 0.9017,
+      "step": 7850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004142391051037054,
+      "loss": 0.7363,
+      "step": 7860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004141225821486833,
+      "loss": 0.7727,
+      "step": 7870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004140060591936611,
+      "loss": 0.7529,
+      "step": 7880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000413889536238639,
+      "loss": 0.8297,
+      "step": 7890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00041377301328361687,
+      "loss": 0.8453,
+      "step": 7900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00041365649032859477,
+      "loss": 0.7607,
+      "step": 7910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00041353996737357256,
+      "loss": 0.787,
+      "step": 7920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00041342344441855047,
+      "loss": 0.8017,
+      "step": 7930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004133069214635283,
+      "loss": 0.6971,
+      "step": 7940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004131903985085062,
+      "loss": 0.8197,
+      "step": 7950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000413073875553484,
+      "loss": 0.7113,
+      "step": 7960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004129573525984619,
+      "loss": 0.871,
+      "step": 7970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00041284082964343976,
+      "loss": 0.7696,
+      "step": 7980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00041272430668841766,
+      "loss": 0.7699,
+      "step": 7990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00041260778373339546,
+      "loss": 0.9623,
+      "step": 8000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00041249126077837336,
+      "loss": 0.7063,
+      "step": 8010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004123747378233512,
+      "loss": 0.811,
+      "step": 8020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004122582148683291,
+      "loss": 0.8056,
+      "step": 8030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004121416919133069,
+      "loss": 0.808,
+      "step": 8040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004120251689582848,
+      "loss": 0.841,
+      "step": 8050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041190864600326266,
+      "loss": 0.7707,
+      "step": 8060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004117921230482405,
+      "loss": 0.7437,
+      "step": 8070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041167560009321835,
+      "loss": 0.8966,
+      "step": 8080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041155907713819626,
+      "loss": 0.7847,
+      "step": 8090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004114425541831741,
+      "loss": 0.7915,
+      "step": 8100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041132603122815195,
+      "loss": 0.8354,
+      "step": 8110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004112095082731298,
+      "loss": 0.7693,
+      "step": 8120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004110929853181077,
+      "loss": 0.7187,
+      "step": 8130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041097646236308555,
+      "loss": 0.7765,
+      "step": 8140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004108599394080634,
+      "loss": 0.8497,
+      "step": 8150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041074341645304125,
+      "loss": 0.7102,
+      "step": 8160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041062689349801915,
+      "loss": 0.746,
+      "step": 8170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000410510370542997,
+      "loss": 0.8032,
+      "step": 8180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00041039384758797485,
+      "loss": 0.8194,
+      "step": 8190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004102773246329527,
+      "loss": 0.8415,
+      "step": 8200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004101608016779306,
+      "loss": 0.7877,
+      "step": 8210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00041004427872290845,
+      "loss": 0.8101,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040992775576788624,
+      "loss": 0.6995,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040981123281286414,
+      "loss": 0.8027,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000409694709857842,
+      "loss": 0.8378,
+      "step": 8250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040957818690281984,
+      "loss": 0.7723,
+      "step": 8260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004094616639477977,
+      "loss": 0.7575,
+      "step": 8270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004093451409927756,
+      "loss": 0.7697,
+      "step": 8280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040922861803775344,
+      "loss": 0.8557,
+      "step": 8290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004091120950827313,
+      "loss": 0.8575,
+      "step": 8300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040899557212770914,
+      "loss": 0.8342,
+      "step": 8310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00040887904917268704,
+      "loss": 0.7827,
+      "step": 8320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0004087625262176649,
+      "loss": 0.822,
+      "step": 8330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00040864600326264274,
+      "loss": 0.7081,
+      "step": 8340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0004085294803076206,
+      "loss": 0.8076,
+      "step": 8350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0004084129573525985,
+      "loss": 0.7991,
+      "step": 8360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040829643439757634,
+      "loss": 0.7308,
+      "step": 8370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0004081799114425542,
+      "loss": 0.8929,
+      "step": 8380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040806338848753203,
+      "loss": 0.7839,
+      "step": 8390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00040794686553250994,
+      "loss": 0.8352,
+      "step": 8400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0004078303425774878,
+      "loss": 0.8239,
+      "step": 8410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00040771381962246563,
+      "loss": 0.9054,
+      "step": 8420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0004075972966674435,
+      "loss": 1.0012,
+      "step": 8430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0004074807737124214,
+      "loss": 0.7988,
+      "step": 8440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0004073642507573992,
+      "loss": 0.7011,
+      "step": 8450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0004072477278023771,
+      "loss": 0.7976,
+      "step": 8460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00040713120484735493,
+      "loss": 0.7668,
+      "step": 8470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00040701468189233283,
+      "loss": 0.887,
+      "step": 8480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0004068981589373106,
+      "loss": 0.7291,
+      "step": 8490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00040678163598228853,
+      "loss": 0.7659,
+      "step": 8500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0004066651130272664,
+      "loss": 0.7434,
+      "step": 8510
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0004065485900722443,
+      "loss": 0.779,
+      "step": 8520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0004064320671172221,
+      "loss": 0.717,
+      "step": 8530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004063155441622,
+      "loss": 0.6918,
+      "step": 8540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004061990212071778,
+      "loss": 0.8615,
+      "step": 8550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004060824982521557,
+      "loss": 0.8585,
+      "step": 8560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004059659752971335,
+      "loss": 0.8218,
+      "step": 8570
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0004058494523421114,
+      "loss": 0.8093,
+      "step": 8580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00040573292938708927,
+      "loss": 0.8097,
+      "step": 8590
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0004056164064320672,
+      "loss": 0.8217,
+      "step": 8600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00040549988347704497,
+      "loss": 0.915,
+      "step": 8610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00040538336052202287,
+      "loss": 0.7121,
+      "step": 8620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004052668375670007,
+      "loss": 0.7469,
+      "step": 8630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00040515031461197857,
+      "loss": 0.7618,
+      "step": 8640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004050337916569564,
+      "loss": 0.7916,
+      "step": 8650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004049172687019343,
+      "loss": 0.7881,
+      "step": 8660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00040480074574691217,
+      "loss": 0.932,
+      "step": 8670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00040468422279189,
+      "loss": 0.8939,
+      "step": 8680
+    },
+    {
+      "epoch": 2.0,
+      "eval_Bleu_1": 0.0061907631622195695,
+      "eval_Bleu_2": 2.2410988520716052e-11,
+      "eval_Bleu_3": 3.913134630236392e-14,
+      "eval_Bleu_4": 1.783965909910145e-15,
+      "eval_ROUGE_L": 0.009656478129255157,
+      "eval_cer": 0.5242234220274556,
+      "eval_em": 0.509952766531714,
+      "eval_f1": 0.6806222594586169,
+      "eval_loss": 0.9325827360153198,
+      "eval_runtime": 333.3233,
+      "eval_samples_per_second": 17.785,
+      "eval_steps_per_second": 3.558,
+      "eval_wer": 0.6251185333473818,
+      "step": 8682
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00040456769983686786,
+      "loss": 0.5991,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0004044511768818457,
+      "loss": 0.5123,
+      "step": 8700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0004043346539268236,
+      "loss": 0.4804,
+      "step": 8710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0004042181309718014,
+      "loss": 0.5649,
+      "step": 8720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0004041016080167793,
+      "loss": 0.5594,
+      "step": 8730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00040398508506175716,
+      "loss": 0.6121,
+      "step": 8740
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00040386856210673506,
+      "loss": 0.4305,
+      "step": 8750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00040375203915171286,
+      "loss": 0.5781,
+      "step": 8760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00040363551619669076,
+      "loss": 0.5346,
+      "step": 8770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0004035189932416686,
+      "loss": 0.5311,
+      "step": 8780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00040340247028664646,
+      "loss": 0.5012,
+      "step": 8790
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0004032859473316243,
+      "loss": 0.5934,
+      "step": 8800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0004031694243766022,
+      "loss": 0.5671,
+      "step": 8810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00040305290142158006,
+      "loss": 0.518,
+      "step": 8820
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0004029363784665579,
+      "loss": 0.552,
+      "step": 8830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00040281985551153575,
+      "loss": 0.5729,
+      "step": 8840
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00040270333255651365,
+      "loss": 0.4851,
+      "step": 8850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0004025868096014915,
+      "loss": 0.4946,
+      "step": 8860
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00040247028664646935,
+      "loss": 0.6723,
+      "step": 8870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0004023537636914472,
+      "loss": 0.4706,
+      "step": 8880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0004022372407364251,
+      "loss": 0.6072,
+      "step": 8890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00040212071778140295,
+      "loss": 0.5382,
+      "step": 8900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0004020041948263808,
+      "loss": 0.5631,
+      "step": 8910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00040188767187135865,
+      "loss": 0.5934,
+      "step": 8920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00040177114891633655,
+      "loss": 0.5487,
+      "step": 8930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0004016546259613144,
+      "loss": 0.5286,
+      "step": 8940
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00040153810300629225,
+      "loss": 0.5221,
+      "step": 8950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0004014215800512701,
+      "loss": 0.527,
+      "step": 8960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000401305057096248,
+      "loss": 0.4899,
+      "step": 8970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0004011885341412258,
+      "loss": 0.5208,
+      "step": 8980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0004010720111862037,
+      "loss": 0.5884,
+      "step": 8990
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00040095548823118154,
+      "loss": 0.5225,
+      "step": 9000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00040083896527615945,
+      "loss": 0.5612,
+      "step": 9010
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00040072244232113724,
+      "loss": 0.6135,
+      "step": 9020
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00040060591936611514,
+      "loss": 0.4754,
+      "step": 9030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000400489396411093,
+      "loss": 0.4943,
+      "step": 9040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0004003728734560709,
+      "loss": 0.5197,
+      "step": 9050
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0004002563505010487,
+      "loss": 0.5642,
+      "step": 9060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0004001398275460266,
+      "loss": 0.5568,
+      "step": 9070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00040002330459100444,
+      "loss": 0.5518,
+      "step": 9080
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00039990678163598234,
+      "loss": 0.6001,
+      "step": 9090
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00039979025868096014,
+      "loss": 0.5839,
+      "step": 9100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00039967373572593804,
+      "loss": 0.6053,
+      "step": 9110
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003995572127709159,
+      "loss": 0.5571,
+      "step": 9120
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003994406898158938,
+      "loss": 0.551,
+      "step": 9130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003993241668608716,
+      "loss": 0.5591,
+      "step": 9140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003992076439058495,
+      "loss": 0.5571,
+      "step": 9150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039909112095082733,
+      "loss": 0.55,
+      "step": 9160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039897459799580513,
+      "loss": 0.6671,
+      "step": 9170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039885807504078303,
+      "loss": 0.6238,
+      "step": 9180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003987415520857609,
+      "loss": 0.4604,
+      "step": 9190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003986250291307388,
+      "loss": 0.576,
+      "step": 9200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003985085061757166,
+      "loss": 0.553,
+      "step": 9210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003983919832206945,
+      "loss": 0.5679,
+      "step": 9220
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003982754602656723,
+      "loss": 0.6115,
+      "step": 9230
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00039815893731065023,
+      "loss": 0.4596,
+      "step": 9240
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000398042414355628,
+      "loss": 0.4852,
+      "step": 9250
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003979258914006059,
+      "loss": 0.5255,
+      "step": 9260
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003978093684455838,
+      "loss": 0.518,
+      "step": 9270
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003976928454905617,
+      "loss": 0.5075,
+      "step": 9280
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00039757632253553947,
+      "loss": 0.6621,
+      "step": 9290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003974597995805174,
+      "loss": 0.5596,
+      "step": 9300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003973432766254952,
+      "loss": 0.4777,
+      "step": 9310
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003972267536704731,
+      "loss": 0.5863,
+      "step": 9320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003971102307154509,
+      "loss": 0.6219,
+      "step": 9330
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003969937077604288,
+      "loss": 0.5444,
+      "step": 9340
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00039687718480540667,
+      "loss": 0.4376,
+      "step": 9350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003967606618503845,
+      "loss": 0.6091,
+      "step": 9360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00039664413889536237,
+      "loss": 0.4934,
+      "step": 9370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00039652761594034027,
+      "loss": 0.5768,
+      "step": 9380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003964110929853181,
+      "loss": 0.5099,
+      "step": 9390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00039629457003029597,
+      "loss": 0.577,
+      "step": 9400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003961780470752738,
+      "loss": 0.5887,
+      "step": 9410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003960615241202517,
+      "loss": 0.5794,
+      "step": 9420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00039594500116522957,
+      "loss": 0.5668,
+      "step": 9430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003958284782102074,
+      "loss": 0.4765,
+      "step": 9440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00039571195525518526,
+      "loss": 0.5262,
+      "step": 9450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00039559543230016317,
+      "loss": 0.6219,
+      "step": 9460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000395478909345141,
+      "loss": 0.5818,
+      "step": 9470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00039536238639011886,
+      "loss": 0.5223,
+      "step": 9480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003952458634350967,
+      "loss": 0.5095,
+      "step": 9490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003951293404800746,
+      "loss": 0.5033,
+      "step": 9500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00039501281752505246,
+      "loss": 0.5708,
+      "step": 9510
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003948962945700303,
+      "loss": 0.5583,
+      "step": 9520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039477977161500816,
+      "loss": 0.5332,
+      "step": 9530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039466324865998606,
+      "loss": 0.6822,
+      "step": 9540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039454672570496385,
+      "loss": 0.4858,
+      "step": 9550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00039443020274994176,
+      "loss": 0.5302,
+      "step": 9560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0003943136797949196,
+      "loss": 0.5078,
+      "step": 9570
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003941971568398975,
+      "loss": 0.5107,
+      "step": 9580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003940806338848753,
+      "loss": 0.5105,
+      "step": 9590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0003939641109298532,
+      "loss": 0.6579,
+      "step": 9600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00039384758797483105,
+      "loss": 0.5097,
+      "step": 9610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00039373106501980896,
+      "loss": 0.575,
+      "step": 9620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00039361454206478675,
+      "loss": 0.6123,
+      "step": 9630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00039349801910976465,
+      "loss": 0.5684,
+      "step": 9640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0003933814961547425,
+      "loss": 0.4895,
+      "step": 9650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003932649731997204,
+      "loss": 0.5427,
+      "step": 9660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003931484502446982,
+      "loss": 0.577,
+      "step": 9670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00039303192728967605,
+      "loss": 0.6597,
+      "step": 9680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00039291540433465395,
+      "loss": 0.5126,
+      "step": 9690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0003927988813796318,
+      "loss": 0.5427,
+      "step": 9700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00039268235842460965,
+      "loss": 0.5729,
+      "step": 9710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003925658354695875,
+      "loss": 0.5166,
+      "step": 9720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003924493125145654,
+      "loss": 0.4773,
+      "step": 9730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0003923327895595432,
+      "loss": 0.6146,
+      "step": 9740
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0003922162666045211,
+      "loss": 0.5842,
+      "step": 9750
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00039209974364949894,
+      "loss": 0.5618,
+      "step": 9760
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00039198322069447684,
+      "loss": 0.5513,
+      "step": 9770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00039186669773945464,
+      "loss": 0.4993,
+      "step": 9780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00039175017478443254,
+      "loss": 0.632,
+      "step": 9790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003916336518294104,
+      "loss": 0.5667,
+      "step": 9800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003915171288743883,
+      "loss": 0.5593,
+      "step": 9810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0003914006059193661,
+      "loss": 0.5384,
+      "step": 9820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000391284082964344,
+      "loss": 0.4804,
+      "step": 9830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00039116756000932184,
+      "loss": 0.5944,
+      "step": 9840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00039105103705429974,
+      "loss": 0.5851,
+      "step": 9850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00039093451409927753,
+      "loss": 0.4737,
+      "step": 9860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00039081799114425544,
+      "loss": 0.645,
+      "step": 9870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003907014681892333,
+      "loss": 0.6055,
+      "step": 9880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003905849452342112,
+      "loss": 0.4633,
+      "step": 9890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000390468422279189,
+      "loss": 0.6038,
+      "step": 9900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0003903518993241669,
+      "loss": 0.5083,
+      "step": 9910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00039023537636914473,
+      "loss": 0.588,
+      "step": 9920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003901188534141226,
+      "loss": 0.6723,
+      "step": 9930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00039000233045910043,
+      "loss": 0.513,
+      "step": 9940
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00038988580750407833,
+      "loss": 0.5506,
+      "step": 9950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0003897692845490562,
+      "loss": 0.595,
+      "step": 9960
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00038965276159403403,
+      "loss": 0.548,
+      "step": 9970
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003895362386390119,
+      "loss": 0.4758,
+      "step": 9980
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0003894197156839898,
+      "loss": 0.5779,
+      "step": 9990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00038930319272896763,
+      "loss": 0.5395,
+      "step": 10000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0003891866697739455,
+      "loss": 0.6145,
+      "step": 10010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0003890701468189233,
+      "loss": 0.598,
+      "step": 10020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00038895362386390123,
+      "loss": 0.6676,
+      "step": 10030
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0003888371009088791,
+      "loss": 0.5585,
+      "step": 10040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003887205779538569,
+      "loss": 0.5398,
+      "step": 10050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00038860405499883477,
+      "loss": 0.5792,
+      "step": 10060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0003884875320438127,
+      "loss": 0.5499,
+      "step": 10070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00038837100908879047,
+      "loss": 0.5817,
+      "step": 10080
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00038825448613376837,
+      "loss": 0.6104,
+      "step": 10090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003881379631787462,
+      "loss": 0.6221,
+      "step": 10100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003880214402237241,
+      "loss": 0.5982,
+      "step": 10110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003879049172687019,
+      "loss": 0.5207,
+      "step": 10120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0003877883943136798,
+      "loss": 0.5528,
+      "step": 10130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00038767187135865767,
+      "loss": 0.5491,
+      "step": 10140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00038755534840363557,
+      "loss": 0.6077,
+      "step": 10150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00038743882544861336,
+      "loss": 0.4959,
+      "step": 10160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003873223024935912,
+      "loss": 0.5356,
+      "step": 10170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0003872057795385691,
+      "loss": 0.5207,
+      "step": 10180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00038708925658354696,
+      "loss": 0.4693,
+      "step": 10190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0003869727336285248,
+      "loss": 0.5714,
+      "step": 10200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00038685621067350266,
+      "loss": 0.5049,
+      "step": 10210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00038673968771848056,
+      "loss": 0.6123,
+      "step": 10220
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003866231647634584,
+      "loss": 0.4514,
+      "step": 10230
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00038650664180843626,
+      "loss": 0.5208,
+      "step": 10240
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0003863901188534141,
+      "loss": 0.6155,
+      "step": 10250
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000386273595898392,
+      "loss": 0.4555,
+      "step": 10260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003861570729433698,
+      "loss": 0.475,
+      "step": 10270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0003860405499883477,
+      "loss": 0.5104,
+      "step": 10280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00038592402703332556,
+      "loss": 0.5335,
+      "step": 10290
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00038580750407830346,
+      "loss": 0.5669,
+      "step": 10300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00038569098112328125,
+      "loss": 0.488,
+      "step": 10310
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00038557445816825916,
+      "loss": 0.5095,
+      "step": 10320
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000385457935213237,
+      "loss": 0.5865,
+      "step": 10330
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003853414122582149,
+      "loss": 0.5239,
+      "step": 10340
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0003852248893031927,
+      "loss": 0.6548,
+      "step": 10350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0003851083663481706,
+      "loss": 0.5929,
+      "step": 10360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00038499184339314845,
+      "loss": 0.5913,
+      "step": 10370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00038487532043812635,
+      "loss": 0.5078,
+      "step": 10380
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00038475879748310415,
+      "loss": 0.5233,
+      "step": 10390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00038464227452808205,
+      "loss": 0.4629,
+      "step": 10400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003845257515730599,
+      "loss": 0.5157,
+      "step": 10410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003844092286180378,
+      "loss": 0.5218,
+      "step": 10420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003842927056630156,
+      "loss": 0.5072,
+      "step": 10430
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0003841761827079935,
+      "loss": 0.6127,
+      "step": 10440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00038405965975297135,
+      "loss": 0.5584,
+      "step": 10450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0003839431367979492,
+      "loss": 0.5054,
+      "step": 10460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00038382661384292704,
+      "loss": 0.5534,
+      "step": 10470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00038371009088790495,
+      "loss": 0.5365,
+      "step": 10480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003835935679328828,
+      "loss": 0.5946,
+      "step": 10490
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00038347704497786064,
+      "loss": 0.6109,
+      "step": 10500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003833605220228385,
+      "loss": 0.6428,
+      "step": 10510
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0003832439990678164,
+      "loss": 0.5929,
+      "step": 10520
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00038312747611279424,
+      "loss": 0.5548,
+      "step": 10530
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0003830109531577721,
+      "loss": 0.5551,
+      "step": 10540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00038289443020274994,
+      "loss": 0.6649,
+      "step": 10550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00038277790724772784,
+      "loss": 0.5881,
+      "step": 10560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0003826613842927057,
+      "loss": 0.6012,
+      "step": 10570
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00038254486133768354,
+      "loss": 0.5753,
+      "step": 10580
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0003824283383826614,
+      "loss": 0.5004,
+      "step": 10590
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0003823118154276393,
+      "loss": 0.6546,
+      "step": 10600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00038219529247261714,
+      "loss": 0.6297,
+      "step": 10610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00038207876951759493,
+      "loss": 0.5892,
+      "step": 10620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00038196224656257284,
+      "loss": 0.5858,
+      "step": 10630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0003818457236075507,
+      "loss": 0.5571,
+      "step": 10640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00038172920065252853,
+      "loss": 0.6245,
+      "step": 10650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0003816126776975064,
+      "loss": 0.5944,
+      "step": 10660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0003814961547424843,
+      "loss": 0.5134,
+      "step": 10670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00038137963178746213,
+      "loss": 0.6271,
+      "step": 10680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00038126310883244,
+      "loss": 0.5594,
+      "step": 10690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00038114658587741783,
+      "loss": 0.6394,
+      "step": 10700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00038103006292239573,
+      "loss": 0.5259,
+      "step": 10710
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0003809135399673736,
+      "loss": 0.5958,
+      "step": 10720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00038079701701235143,
+      "loss": 0.6885,
+      "step": 10730
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0003806804940573293,
+      "loss": 0.5735,
+      "step": 10740
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0003805639711023072,
+      "loss": 0.5948,
+      "step": 10750
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000380447448147285,
+      "loss": 0.4641,
+      "step": 10760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0003803309251922629,
+      "loss": 0.5771,
+      "step": 10770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0003802144022372407,
+      "loss": 0.5432,
+      "step": 10780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003800978792822186,
+      "loss": 0.5765,
+      "step": 10790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003799813563271965,
+      "loss": 0.6005,
+      "step": 10800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003798648333721743,
+      "loss": 0.4986,
+      "step": 10810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00037974831041715217,
+      "loss": 0.5309,
+      "step": 10820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0003796317874621301,
+      "loss": 0.5259,
+      "step": 10830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00037951526450710787,
+      "loss": 0.4959,
+      "step": 10840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00037939874155208577,
+      "loss": 0.5123,
+      "step": 10850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003792822185970636,
+      "loss": 0.5946,
+      "step": 10860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0003791656956420415,
+      "loss": 0.5913,
+      "step": 10870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003790491726870193,
+      "loss": 0.5857,
+      "step": 10880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0003789326497319972,
+      "loss": 0.5496,
+      "step": 10890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00037881612677697507,
+      "loss": 0.5672,
+      "step": 10900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00037869960382195297,
+      "loss": 0.5571,
+      "step": 10910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00037858308086693076,
+      "loss": 0.5366,
+      "step": 10920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00037846655791190867,
+      "loss": 0.5317,
+      "step": 10930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003783500349568865,
+      "loss": 0.506,
+      "step": 10940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003782335120018644,
+      "loss": 0.6632,
+      "step": 10950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0003781169890468422,
+      "loss": 0.548,
+      "step": 10960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0003780004660918201,
+      "loss": 0.4903,
+      "step": 10970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00037788394313679796,
+      "loss": 0.4934,
+      "step": 10980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00037776742018177586,
+      "loss": 0.5766,
+      "step": 10990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00037765089722675366,
+      "loss": 0.5807,
+      "step": 11000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00037753437427173156,
+      "loss": 0.537,
+      "step": 11010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003774178513167094,
+      "loss": 0.4981,
+      "step": 11020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00037730132836168726,
+      "loss": 0.6566,
+      "step": 11030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0003771848054066651,
+      "loss": 0.5681,
+      "step": 11040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000377068282451643,
+      "loss": 0.5734,
+      "step": 11050
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00037695175949662086,
+      "loss": 0.4979,
+      "step": 11060
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0003768352365415987,
+      "loss": 0.5895,
+      "step": 11070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00037671871358657655,
+      "loss": 0.6714,
+      "step": 11080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00037660219063155446,
+      "loss": 0.66,
+      "step": 11090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003764856676765323,
+      "loss": 0.5396,
+      "step": 11100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0003763691447215101,
+      "loss": 0.5783,
+      "step": 11110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000376252621766488,
+      "loss": 0.5514,
+      "step": 11120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00037613609881146585,
+      "loss": 0.7365,
+      "step": 11130
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00037601957585644375,
+      "loss": 0.602,
+      "step": 11140
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00037590305290142155,
+      "loss": 0.7115,
+      "step": 11150
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00037578652994639945,
+      "loss": 0.5712,
+      "step": 11160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0003756700069913773,
+      "loss": 0.6067,
+      "step": 11170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0003755534840363552,
+      "loss": 0.5876,
+      "step": 11180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000375436961081333,
+      "loss": 0.5376,
+      "step": 11190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0003753204381263109,
+      "loss": 0.686,
+      "step": 11200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00037520391517128875,
+      "loss": 0.511,
+      "step": 11210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0003750873922162666,
+      "loss": 0.5709,
+      "step": 11220
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00037497086926124444,
+      "loss": 0.5382,
+      "step": 11230
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00037485434630622235,
+      "loss": 0.6335,
+      "step": 11240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0003747378233512002,
+      "loss": 0.6076,
+      "step": 11250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00037462130039617804,
+      "loss": 0.741,
+      "step": 11260
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003745047774411559,
+      "loss": 0.6341,
+      "step": 11270
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003743882544861338,
+      "loss": 0.5657,
+      "step": 11280
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00037427173153111164,
+      "loss": 0.6267,
+      "step": 11290
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0003741552085760895,
+      "loss": 0.6451,
+      "step": 11300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00037403868562106734,
+      "loss": 0.5806,
+      "step": 11310
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00037392216266604524,
+      "loss": 0.6063,
+      "step": 11320
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003738056397110231,
+      "loss": 0.6284,
+      "step": 11330
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00037368911675600094,
+      "loss": 0.4528,
+      "step": 11340
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0003735725938009788,
+      "loss": 0.6778,
+      "step": 11350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003734560708459567,
+      "loss": 0.6654,
+      "step": 11360
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003733395478909345,
+      "loss": 0.5415,
+      "step": 11370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0003732230249359124,
+      "loss": 0.537,
+      "step": 11380
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00037310650198089023,
+      "loss": 0.5016,
+      "step": 11390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00037298997902586814,
+      "loss": 0.5729,
+      "step": 11400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00037287345607084593,
+      "loss": 0.5655,
+      "step": 11410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00037275693311582383,
+      "loss": 0.7096,
+      "step": 11420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0003726404101608017,
+      "loss": 0.6308,
+      "step": 11430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003725238872057796,
+      "loss": 0.505,
+      "step": 11440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003724073642507574,
+      "loss": 0.6018,
+      "step": 11450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0003722908412957353,
+      "loss": 0.5669,
+      "step": 11460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00037217431834071313,
+      "loss": 0.6607,
+      "step": 11470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00037205779538569103,
+      "loss": 0.4848,
+      "step": 11480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003719412724306688,
+      "loss": 0.6305,
+      "step": 11490
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00037182474947564673,
+      "loss": 0.6011,
+      "step": 11500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003717082265206246,
+      "loss": 0.5572,
+      "step": 11510
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0003715917035656025,
+      "loss": 0.5913,
+      "step": 11520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003714751806105803,
+      "loss": 0.6331,
+      "step": 11530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003713586576555582,
+      "loss": 0.5433,
+      "step": 11540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000371242134700536,
+      "loss": 0.5089,
+      "step": 11550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0003711256117455139,
+      "loss": 0.4861,
+      "step": 11560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0003710090887904917,
+      "loss": 0.602,
+      "step": 11570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0003708925658354696,
+      "loss": 0.4407,
+      "step": 11580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00037077604288044747,
+      "loss": 0.6049,
+      "step": 11590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00037065951992542527,
+      "loss": 0.6089,
+      "step": 11600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00037054299697040317,
+      "loss": 0.4972,
+      "step": 11610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000370426474015381,
+      "loss": 0.5839,
+      "step": 11620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003703099510603589,
+      "loss": 0.5832,
+      "step": 11630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003701934281053367,
+      "loss": 0.5513,
+      "step": 11640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0003700769051503146,
+      "loss": 0.5815,
+      "step": 11650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00036996038219529247,
+      "loss": 0.6414,
+      "step": 11660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00036984385924027037,
+      "loss": 0.4812,
+      "step": 11670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00036972733628524816,
+      "loss": 0.5898,
+      "step": 11680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00036961081333022606,
+      "loss": 0.6138,
+      "step": 11690
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003694942903752039,
+      "loss": 0.5094,
+      "step": 11700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003693777674201818,
+      "loss": 0.6374,
+      "step": 11710
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003692612444651596,
+      "loss": 0.5514,
+      "step": 11720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0003691447215101375,
+      "loss": 0.4668,
+      "step": 11730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00036902819855511536,
+      "loss": 0.5207,
+      "step": 11740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0003689116756000932,
+      "loss": 0.5768,
+      "step": 11750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00036879515264507106,
+      "loss": 0.5295,
+      "step": 11760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00036867862969004896,
+      "loss": 0.5756,
+      "step": 11770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0003685621067350268,
+      "loss": 0.5905,
+      "step": 11780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00036844558378000466,
+      "loss": 0.5398,
+      "step": 11790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0003683290608249825,
+      "loss": 0.5105,
+      "step": 11800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0003682125378699604,
+      "loss": 0.449,
+      "step": 11810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00036809601491493826,
+      "loss": 0.6313,
+      "step": 11820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003679794919599161,
+      "loss": 0.5288,
+      "step": 11830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00036786296900489395,
+      "loss": 0.5334,
+      "step": 11840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00036774644604987186,
+      "loss": 0.5581,
+      "step": 11850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0003676299230948497,
+      "loss": 0.5738,
+      "step": 11860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00036751340013982755,
+      "loss": 0.5783,
+      "step": 11870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003673968771848054,
+      "loss": 0.5406,
+      "step": 11880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0003672803542297833,
+      "loss": 0.6214,
+      "step": 11890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00036716383127476115,
+      "loss": 0.5412,
+      "step": 11900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000367047308319739,
+      "loss": 0.6317,
+      "step": 11910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036693078536471685,
+      "loss": 0.617,
+      "step": 11920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036681426240969475,
+      "loss": 0.5979,
+      "step": 11930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036669773945467255,
+      "loss": 0.5684,
+      "step": 11940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00036658121649965045,
+      "loss": 0.632,
+      "step": 11950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0003664646935446283,
+      "loss": 0.544,
+      "step": 11960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003663481705896062,
+      "loss": 0.591,
+      "step": 11970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000366231647634584,
+      "loss": 0.5811,
+      "step": 11980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0003661151246795619,
+      "loss": 0.5066,
+      "step": 11990
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00036599860172453974,
+      "loss": 0.5227,
+      "step": 12000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00036588207876951765,
+      "loss": 0.6378,
+      "step": 12010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00036576555581449544,
+      "loss": 0.5862,
+      "step": 12020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00036564903285947334,
+      "loss": 0.6071,
+      "step": 12030
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0003655325099044512,
+      "loss": 0.5584,
+      "step": 12040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003654159869494291,
+      "loss": 0.6494,
+      "step": 12050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003652994639944069,
+      "loss": 0.5216,
+      "step": 12060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003651829410393848,
+      "loss": 0.5979,
+      "step": 12070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00036506641808436264,
+      "loss": 0.5455,
+      "step": 12080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0003649498951293405,
+      "loss": 0.6383,
+      "step": 12090
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00036483337217431834,
+      "loss": 0.4883,
+      "step": 12100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003647168492192962,
+      "loss": 0.6385,
+      "step": 12110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003646003262642741,
+      "loss": 0.6161,
+      "step": 12120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0003644838033092519,
+      "loss": 0.6531,
+      "step": 12130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0003643672803542298,
+      "loss": 0.7073,
+      "step": 12140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00036425075739920763,
+      "loss": 0.6292,
+      "step": 12150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00036413423444418553,
+      "loss": 0.532,
+      "step": 12160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00036401771148916333,
+      "loss": 0.5598,
+      "step": 12170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00036390118853414123,
+      "loss": 0.5761,
+      "step": 12180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003637846655791191,
+      "loss": 0.6883,
+      "step": 12190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000363668142624097,
+      "loss": 0.531,
+      "step": 12200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003635516196690748,
+      "loss": 0.5152,
+      "step": 12210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0003634350967140527,
+      "loss": 0.4439,
+      "step": 12220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00036331857375903053,
+      "loss": 0.6116,
+      "step": 12230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00036320205080400843,
+      "loss": 0.575,
+      "step": 12240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0003630855278489862,
+      "loss": 0.6519,
+      "step": 12250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00036296900489396413,
+      "loss": 0.7286,
+      "step": 12260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000362852481938942,
+      "loss": 0.6153,
+      "step": 12270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003627359589839199,
+      "loss": 0.6239,
+      "step": 12280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00036261943602889767,
+      "loss": 0.5667,
+      "step": 12290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0003625029130738756,
+      "loss": 0.5172,
+      "step": 12300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003623863901188534,
+      "loss": 0.536,
+      "step": 12310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00036226986716383127,
+      "loss": 0.4811,
+      "step": 12320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0003621533442088091,
+      "loss": 0.5181,
+      "step": 12330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000362036821253787,
+      "loss": 0.6049,
+      "step": 12340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00036192029829876487,
+      "loss": 0.6295,
+      "step": 12350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0003618037753437427,
+      "loss": 0.5561,
+      "step": 12360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00036168725238872057,
+      "loss": 0.5883,
+      "step": 12370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00036157072943369847,
+      "loss": 0.5432,
+      "step": 12380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0003614542064786763,
+      "loss": 0.4879,
+      "step": 12390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00036133768352365417,
+      "loss": 0.5342,
+      "step": 12400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000361221160568632,
+      "loss": 0.6041,
+      "step": 12410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0003611046376136099,
+      "loss": 0.5651,
+      "step": 12420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00036098811465858777,
+      "loss": 0.5147,
+      "step": 12430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003608715917035656,
+      "loss": 0.5625,
+      "step": 12440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00036075506874854346,
+      "loss": 0.5852,
+      "step": 12450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00036063854579352137,
+      "loss": 0.5484,
+      "step": 12460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0003605220228384992,
+      "loss": 0.5484,
+      "step": 12470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00036040549988347706,
+      "loss": 0.7071,
+      "step": 12480
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003602889769284549,
+      "loss": 0.5853,
+      "step": 12490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003601724539734328,
+      "loss": 0.5038,
+      "step": 12500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003600559310184106,
+      "loss": 0.5228,
+      "step": 12510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0003599394080633885,
+      "loss": 0.6229,
+      "step": 12520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00035982288510836636,
+      "loss": 0.6315,
+      "step": 12530
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00035970636215334426,
+      "loss": 0.5097,
+      "step": 12540
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00035958983919832206,
+      "loss": 0.5466,
+      "step": 12550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0003594733162432999,
+      "loss": 0.5656,
+      "step": 12560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0003593567932882778,
+      "loss": 0.6161,
+      "step": 12570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00035924027033325565,
+      "loss": 0.5725,
+      "step": 12580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0003591237473782335,
+      "loss": 0.5583,
+      "step": 12590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00035900722442321135,
+      "loss": 0.5669,
+      "step": 12600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00035889070146818925,
+      "loss": 0.5418,
+      "step": 12610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003587741785131671,
+      "loss": 0.5785,
+      "step": 12620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00035865765555814495,
+      "loss": 0.5889,
+      "step": 12630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003585411326031228,
+      "loss": 0.5348,
+      "step": 12640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0003584246096481007,
+      "loss": 0.5271,
+      "step": 12650
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003583080866930785,
+      "loss": 0.5257,
+      "step": 12660
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0003581915637380564,
+      "loss": 0.571,
+      "step": 12670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00035807504078303425,
+      "loss": 0.553,
+      "step": 12680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00035795851782801215,
+      "loss": 0.5103,
+      "step": 12690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00035784199487298994,
+      "loss": 0.5788,
+      "step": 12700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00035772547191796785,
+      "loss": 0.5475,
+      "step": 12710
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003576089489629457,
+      "loss": 0.5787,
+      "step": 12720
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003574924260079236,
+      "loss": 0.6733,
+      "step": 12730
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0003573759030529014,
+      "loss": 0.5088,
+      "step": 12740
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0003572593800978793,
+      "loss": 0.5526,
+      "step": 12750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00035714285714285714,
+      "loss": 0.5531,
+      "step": 12760
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00035702633418783505,
+      "loss": 0.5685,
+      "step": 12770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00035690981123281284,
+      "loss": 0.5849,
+      "step": 12780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00035679328827779074,
+      "loss": 0.598,
+      "step": 12790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003566767653227686,
+      "loss": 0.6349,
+      "step": 12800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003565602423677465,
+      "loss": 0.5224,
+      "step": 12810
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0003564437194127243,
+      "loss": 0.5802,
+      "step": 12820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003563271964577022,
+      "loss": 0.6304,
+      "step": 12830
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00035621067350268004,
+      "loss": 0.6912,
+      "step": 12840
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0003560941505476579,
+      "loss": 0.5772,
+      "step": 12850
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00035597762759263573,
+      "loss": 0.5425,
+      "step": 12860
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00035586110463761364,
+      "loss": 0.6595,
+      "step": 12870
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003557445816825915,
+      "loss": 0.5792,
+      "step": 12880
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00035562805872756933,
+      "loss": 0.5013,
+      "step": 12890
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003555115357725472,
+      "loss": 0.7614,
+      "step": 12900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0003553950128175251,
+      "loss": 0.5491,
+      "step": 12910
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00035527848986250293,
+      "loss": 0.5925,
+      "step": 12920
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0003551619669074808,
+      "loss": 0.4564,
+      "step": 12930
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00035504544395245863,
+      "loss": 0.5478,
+      "step": 12940
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00035492892099743653,
+      "loss": 0.567,
+      "step": 12950
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003548123980424144,
+      "loss": 0.6399,
+      "step": 12960
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00035469587508739223,
+      "loss": 0.5897,
+      "step": 12970
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0003545793521323701,
+      "loss": 0.5591,
+      "step": 12980
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000354462829177348,
+      "loss": 0.597,
+      "step": 12990
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00035434630622232583,
+      "loss": 0.5599,
+      "step": 13000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003542297832673037,
+      "loss": 0.7182,
+      "step": 13010
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003541132603122815,
+      "loss": 0.6147,
+      "step": 13020
+    },
+    {
+      "epoch": 3.0,
+      "eval_Bleu_1": 0.004867795110078279,
+      "eval_Bleu_2": 2.0016032894805002e-11,
+      "eval_Bleu_3": 3.641044386384566e-14,
+      "eval_Bleu_4": 1.688390378855272e-15,
+      "eval_ROUGE_L": 0.0076157806660195025,
+      "eval_cer": 0.4872366010539993,
+      "eval_em": 0.5393049932523617,
+      "eval_f1": 0.7065047406756789,
+      "eval_loss": 0.8688169717788696,
+      "eval_runtime": 329.7388,
+      "eval_samples_per_second": 17.978,
+      "eval_steps_per_second": 3.597,
+      "eval_wer": 0.5752291644716047,
+      "step": 13023
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00035399673735725943,
+      "loss": 0.4246,
+      "step": 13030
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003538802144022372,
+      "loss": 0.3049,
+      "step": 13040
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00035376369144721507,
+      "loss": 0.3106,
+      "step": 13050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000353647168492193,
+      "loss": 0.3329,
+      "step": 13060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003535306455371708,
+      "loss": 0.481,
+      "step": 13070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00035341412258214867,
+      "loss": 0.3728,
+      "step": 13080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003532975996271265,
+      "loss": 0.3044,
+      "step": 13090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003531810766721044,
+      "loss": 0.4434,
+      "step": 13100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00035306455371708227,
+      "loss": 0.397,
+      "step": 13110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003529480307620601,
+      "loss": 0.4271,
+      "step": 13120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00035283150780703797,
+      "loss": 0.3274,
+      "step": 13130
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00035271498485201587,
+      "loss": 0.3705,
+      "step": 13140
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003525984618969937,
+      "loss": 0.3008,
+      "step": 13150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00035248193894197157,
+      "loss": 0.3671,
+      "step": 13160
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0003523654159869494,
+      "loss": 0.3638,
+      "step": 13170
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003522488930319273,
+      "loss": 0.2577,
+      "step": 13180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00035213237007690517,
+      "loss": 0.3867,
+      "step": 13190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000352015847121883,
+      "loss": 0.411,
+      "step": 13200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00035189932416686086,
+      "loss": 0.3793,
+      "step": 13210
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00035178280121183876,
+      "loss": 0.3405,
+      "step": 13220
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00035166627825681656,
+      "loss": 0.3649,
+      "step": 13230
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00035154975530179446,
+      "loss": 0.3443,
+      "step": 13240
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003514332323467723,
+      "loss": 0.467,
+      "step": 13250
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003513167093917502,
+      "loss": 0.3444,
+      "step": 13260
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000351200186436728,
+      "loss": 0.3961,
+      "step": 13270
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003510836634817059,
+      "loss": 0.3478,
+      "step": 13280
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00035096714052668376,
+      "loss": 0.3768,
+      "step": 13290
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00035085061757166166,
+      "loss": 0.3578,
+      "step": 13300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00035073409461663945,
+      "loss": 0.3998,
+      "step": 13310
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00035061757166161736,
+      "loss": 0.4058,
+      "step": 13320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003505010487065952,
+      "loss": 0.3761,
+      "step": 13330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003503845257515731,
+      "loss": 0.3347,
+      "step": 13340
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003502680027965509,
+      "loss": 0.3976,
+      "step": 13350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0003501514798415288,
+      "loss": 0.3189,
+      "step": 13360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00035003495688650665,
+      "loss": 0.302,
+      "step": 13370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00034991843393148456,
+      "loss": 0.3786,
+      "step": 13380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00034980191097646235,
+      "loss": 0.4283,
+      "step": 13390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00034968538802144025,
+      "loss": 0.4397,
+      "step": 13400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003495688650664181,
+      "loss": 0.3887,
+      "step": 13410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00034945234211139595,
+      "loss": 0.3179,
+      "step": 13420
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003493358191563738,
+      "loss": 0.3351,
+      "step": 13430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0003492192962013517,
+      "loss": 0.3515,
+      "step": 13440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00034910277324632955,
+      "loss": 0.3272,
+      "step": 13450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0003489862502913074,
+      "loss": 0.3358,
+      "step": 13460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00034886972733628525,
+      "loss": 0.3854,
+      "step": 13470
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00034875320438126315,
+      "loss": 0.3919,
+      "step": 13480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000348636681426241,
+      "loss": 0.2705,
+      "step": 13490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00034852015847121884,
+      "loss": 0.3102,
+      "step": 13500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003484036355161967,
+      "loss": 0.5056,
+      "step": 13510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003482871125611746,
+      "loss": 0.3482,
+      "step": 13520
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00034817058960615244,
+      "loss": 0.3364,
+      "step": 13530
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00034805406665113024,
+      "loss": 0.328,
+      "step": 13540
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00034793754369610814,
+      "loss": 0.3484,
+      "step": 13550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000347821020741086,
+      "loss": 0.3948,
+      "step": 13560
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0003477044977860639,
+      "loss": 0.3593,
+      "step": 13570
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0003475879748310417,
+      "loss": 0.3694,
+      "step": 13580
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0003474714518760196,
+      "loss": 0.3409,
+      "step": 13590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00034735492892099744,
+      "loss": 0.4645,
+      "step": 13600
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003472384059659753,
+      "loss": 0.3569,
+      "step": 13610
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00034712188301095313,
+      "loss": 0.3359,
+      "step": 13620
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00034700536005593104,
+      "loss": 0.3948,
+      "step": 13630
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003468888371009089,
+      "loss": 0.3631,
+      "step": 13640
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00034677231414588673,
+      "loss": 0.3676,
+      "step": 13650
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003466557911908646,
+      "loss": 0.3725,
+      "step": 13660
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003465392682358425,
+      "loss": 0.3383,
+      "step": 13670
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00034642274528082033,
+      "loss": 0.4721,
+      "step": 13680
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003463062223257982,
+      "loss": 0.3461,
+      "step": 13690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00034618969937077603,
+      "loss": 0.3888,
+      "step": 13700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00034607317641575393,
+      "loss": 0.3821,
+      "step": 13710
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003459566534607318,
+      "loss": 0.3944,
+      "step": 13720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00034584013050570963,
+      "loss": 0.41,
+      "step": 13730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003457236075506875,
+      "loss": 0.5004,
+      "step": 13740
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003456070845956654,
+      "loss": 0.3916,
+      "step": 13750
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003454905616406432,
+      "loss": 0.4201,
+      "step": 13760
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003453740386856211,
+      "loss": 0.3571,
+      "step": 13770
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0003452575157305989,
+      "loss": 0.3769,
+      "step": 13780
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003451409927755768,
+      "loss": 0.4201,
+      "step": 13790
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003450244698205546,
+      "loss": 0.4101,
+      "step": 13800
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003449079468655325,
+      "loss": 0.3391,
+      "step": 13810
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00034479142391051037,
+      "loss": 0.4024,
+      "step": 13820
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0003446749009554883,
+      "loss": 0.3296,
+      "step": 13830
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00034455837800046607,
+      "loss": 0.3834,
+      "step": 13840
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00034444185504544397,
+      "loss": 0.3787,
+      "step": 13850
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0003443253320904218,
+      "loss": 0.481,
+      "step": 13860
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0003442088091353997,
+      "loss": 0.3975,
+      "step": 13870
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0003440922861803775,
+      "loss": 0.368,
+      "step": 13880
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0003439757632253554,
+      "loss": 0.3682,
+      "step": 13890
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00034385924027033327,
+      "loss": 0.3515,
+      "step": 13900
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00034374271731531117,
+      "loss": 0.3828,
+      "step": 13910
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00034362619436028896,
+      "loss": 0.315,
+      "step": 13920
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00034350967140526687,
+      "loss": 0.385,
+      "step": 13930
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0003433931484502447,
+      "loss": 0.2604,
+      "step": 13940
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00034327662549522256,
+      "loss": 0.319,
+      "step": 13950
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0003431601025402004,
+      "loss": 0.3606,
+      "step": 13960
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0003430435795851783,
+      "loss": 0.3567,
+      "step": 13970
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00034292705663015616,
+      "loss": 0.4613,
+      "step": 13980
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00034281053367513396,
+      "loss": 0.385,
+      "step": 13990
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00034269401072011186,
+      "loss": 0.3596,
+      "step": 14000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003425774877650897,
+      "loss": 0.3266,
+      "step": 14010
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003424609648100676,
+      "loss": 0.3339,
+      "step": 14020
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003423444418550454,
+      "loss": 0.4464,
+      "step": 14030
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003422279189000233,
+      "loss": 0.3633,
+      "step": 14040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034211139594500116,
+      "loss": 0.4304,
+      "step": 14050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034199487298997906,
+      "loss": 0.3784,
+      "step": 14060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034187835003495685,
+      "loss": 0.2993,
+      "step": 14070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00034176182707993476,
+      "loss": 0.3642,
+      "step": 14080
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003416453041249126,
+      "loss": 0.3952,
+      "step": 14090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003415287811698905,
+      "loss": 0.3625,
+      "step": 14100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003414122582148683,
+      "loss": 0.4257,
+      "step": 14110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003412957352598462,
+      "loss": 0.362,
+      "step": 14120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00034117921230482405,
+      "loss": 0.3809,
+      "step": 14130
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0003410626893498019,
+      "loss": 0.4442,
+      "step": 14140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00034094616639477975,
+      "loss": 0.3999,
+      "step": 14150
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00034082964343975765,
+      "loss": 0.3906,
+      "step": 14160
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0003407131204847355,
+      "loss": 0.3515,
+      "step": 14170
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00034059659752971335,
+      "loss": 0.3556,
+      "step": 14180
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0003404800745746912,
+      "loss": 0.3242,
+      "step": 14190
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0003403635516196691,
+      "loss": 0.4274,
+      "step": 14200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00034024702866464695,
+      "loss": 0.4594,
+      "step": 14210
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0003401305057096248,
+      "loss": 0.4227,
+      "step": 14220
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00034001398275460264,
+      "loss": 0.358,
+      "step": 14230
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00033989745979958055,
+      "loss": 0.3896,
+      "step": 14240
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0003397809368445584,
+      "loss": 0.3688,
+      "step": 14250
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00033966441388953624,
+      "loss": 0.3999,
+      "step": 14260
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0003395478909345141,
+      "loss": 0.4164,
+      "step": 14270
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000339431367979492,
+      "loss": 0.43,
+      "step": 14280
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00033931484502446984,
+      "loss": 0.3852,
+      "step": 14290
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0003391983220694477,
+      "loss": 0.3611,
+      "step": 14300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00033908179911442554,
+      "loss": 0.4676,
+      "step": 14310
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00033896527615940344,
+      "loss": 0.3762,
+      "step": 14320
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00033884875320438124,
+      "loss": 0.4117,
+      "step": 14330
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00033873223024935914,
+      "loss": 0.3538,
+      "step": 14340
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000338615707294337,
+      "loss": 0.4657,
+      "step": 14350
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0003384991843393149,
+      "loss": 0.352,
+      "step": 14360
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0003383826613842927,
+      "loss": 0.3408,
+      "step": 14370
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0003382661384292706,
+      "loss": 0.3885,
+      "step": 14380
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00033814961547424843,
+      "loss": 0.3607,
+      "step": 14390
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00033803309251922634,
+      "loss": 0.4017,
+      "step": 14400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00033791656956420413,
+      "loss": 0.3497,
+      "step": 14410
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00033780004660918203,
+      "loss": 0.3996,
+      "step": 14420
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0003376835236541599,
+      "loss": 0.4579,
+      "step": 14430
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003375670006991378,
+      "loss": 0.374,
+      "step": 14440
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003374504777441156,
+      "loss": 0.3747,
+      "step": 14450
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003373339547890935,
+      "loss": 0.4304,
+      "step": 14460
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00033721743183407133,
+      "loss": 0.4045,
+      "step": 14470
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00033710090887904923,
+      "loss": 0.3762,
+      "step": 14480
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000336984385924027,
+      "loss": 0.4466,
+      "step": 14490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0003368678629690049,
+      "loss": 0.4148,
+      "step": 14500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0003367513400139828,
+      "loss": 0.3531,
+      "step": 14510
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00033663481705896057,
+      "loss": 0.3318,
+      "step": 14520
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0003365182941039385,
+      "loss": 0.3931,
+      "step": 14530
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0003364017711489163,
+      "loss": 0.2862,
+      "step": 14540
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0003362852481938942,
+      "loss": 0.4828,
+      "step": 14550
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.000336168725238872,
+      "loss": 0.3579,
+      "step": 14560
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0003360522022838499,
+      "loss": 0.346,
+      "step": 14570
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00033593567932882777,
+      "loss": 0.4179,
+      "step": 14580
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0003358191563738057,
+      "loss": 0.4223,
+      "step": 14590
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00033570263341878347,
+      "loss": 0.4155,
+      "step": 14600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00033558611046376137,
+      "loss": 0.3151,
+      "step": 14610
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0003354695875087392,
+      "loss": 0.3775,
+      "step": 14620
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0003353530645537171,
+      "loss": 0.4089,
+      "step": 14630
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0003352365415986949,
+      "loss": 0.3432,
+      "step": 14640
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0003351200186436728,
+      "loss": 0.3481,
+      "step": 14650
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033500349568865067,
+      "loss": 0.332,
+      "step": 14660
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033488697273362857,
+      "loss": 0.3366,
+      "step": 14670
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033477044977860636,
+      "loss": 0.5104,
+      "step": 14680
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00033465392682358427,
+      "loss": 0.4099,
+      "step": 14690
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0003345374038685621,
+      "loss": 0.3269,
+      "step": 14700
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00033442088091353996,
+      "loss": 0.3343,
+      "step": 14710
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0003343043579585178,
+      "loss": 0.4143,
+      "step": 14720
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0003341878350034957,
+      "loss": 0.3872,
+      "step": 14730
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033407131204847356,
+      "loss": 0.3662,
+      "step": 14740
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0003339547890934514,
+      "loss": 0.3666,
+      "step": 14750
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033383826613842926,
+      "loss": 0.4467,
+      "step": 14760
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033372174318340716,
+      "loss": 0.4086,
+      "step": 14770
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000333605220228385,
+      "loss": 0.4165,
+      "step": 14780
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00033348869727336286,
+      "loss": 0.3612,
+      "step": 14790
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0003333721743183407,
+      "loss": 0.3583,
+      "step": 14800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0003332556513633186,
+      "loss": 0.3849,
+      "step": 14810
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00033313912840829646,
+      "loss": 0.4263,
+      "step": 14820
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0003330226054532743,
+      "loss": 0.2838,
+      "step": 14830
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00033290608249825215,
+      "loss": 0.2699,
+      "step": 14840
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00033278955954323006,
+      "loss": 0.3844,
+      "step": 14850
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0003326730365882079,
+      "loss": 0.3721,
+      "step": 14860
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00033255651363318575,
+      "loss": 0.4008,
+      "step": 14870
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003324399906781636,
+      "loss": 0.4276,
+      "step": 14880
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003323234677231415,
+      "loss": 0.3757,
+      "step": 14890
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003322069447681193,
+      "loss": 0.4514,
+      "step": 14900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0003320904218130972,
+      "loss": 0.3552,
+      "step": 14910
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00033197389885807505,
+      "loss": 0.4417,
+      "step": 14920
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00033185737590305295,
+      "loss": 0.3657,
+      "step": 14930
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00033174085294803075,
+      "loss": 0.3678,
+      "step": 14940
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00033162432999300865,
+      "loss": 0.3587,
+      "step": 14950
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0003315078070379865,
+      "loss": 0.3822,
+      "step": 14960
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0003313912840829644,
+      "loss": 0.4112,
+      "step": 14970
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0003312747611279422,
+      "loss": 0.4425,
+      "step": 14980
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00033115823817292004,
+      "loss": 0.4285,
+      "step": 14990
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00033104171521789794,
+      "loss": 0.4583,
+      "step": 15000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0003309251922628758,
+      "loss": 0.497,
+      "step": 15010
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00033080866930785364,
+      "loss": 0.3621,
+      "step": 15020
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0003306921463528315,
+      "loss": 0.3378,
+      "step": 15030
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0003305756233978094,
+      "loss": 0.3557,
+      "step": 15040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0003304591004427872,
+      "loss": 0.3905,
+      "step": 15050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0003303425774877651,
+      "loss": 0.417,
+      "step": 15060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00033022605453274294,
+      "loss": 0.4327,
+      "step": 15070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00033010953157772084,
+      "loss": 0.4692,
+      "step": 15080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00032999300862269863,
+      "loss": 0.3902,
+      "step": 15090
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00032987648566767654,
+      "loss": 0.3913,
+      "step": 15100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003297599627126544,
+      "loss": 0.3976,
+      "step": 15110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003296434397576323,
+      "loss": 0.3878,
+      "step": 15120
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0003295269168026101,
+      "loss": 0.5335,
+      "step": 15130
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000329410393847588,
+      "loss": 0.3805,
+      "step": 15140
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00032929387089256583,
+      "loss": 0.4133,
+      "step": 15150
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00032917734793754374,
+      "loss": 0.3966,
+      "step": 15160
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00032906082498252153,
+      "loss": 0.3819,
+      "step": 15170
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00032894430202749943,
+      "loss": 0.3684,
+      "step": 15180
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0003288277790724773,
+      "loss": 0.3883,
+      "step": 15190
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0003287112561174552,
+      "loss": 0.3613,
+      "step": 15200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000328594733162433,
+      "loss": 0.4092,
+      "step": 15210
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0003284782102074109,
+      "loss": 0.3864,
+      "step": 15220
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00032836168725238873,
+      "loss": 0.3888,
+      "step": 15230
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0003282451642973666,
+      "loss": 0.4062,
+      "step": 15240
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0003281286413423444,
+      "loss": 0.4343,
+      "step": 15250
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00032801211838732233,
+      "loss": 0.4947,
+      "step": 15260
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0003278955954323002,
+      "loss": 0.3297,
+      "step": 15270
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.000327779072477278,
+      "loss": 0.3375,
+      "step": 15280
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0003276625495222559,
+      "loss": 0.3941,
+      "step": 15290
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0003275460265672338,
+      "loss": 0.4277,
+      "step": 15300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0003274295036122116,
+      "loss": 0.426,
+      "step": 15310
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00032731298065718947,
+      "loss": 0.3812,
+      "step": 15320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0003271964577021673,
+      "loss": 0.4415,
+      "step": 15330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0003270799347471452,
+      "loss": 0.3589,
+      "step": 15340
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00032696341179212307,
+      "loss": 0.4216,
+      "step": 15350
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0003268468888371009,
+      "loss": 0.4259,
+      "step": 15360
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00032673036588207877,
+      "loss": 0.4129,
+      "step": 15370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00032661384292705667,
+      "loss": 0.4317,
+      "step": 15380
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0003264973199720345,
+      "loss": 0.4966,
+      "step": 15390
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00032638079701701237,
+      "loss": 0.4386,
+      "step": 15400
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0003262642740619902,
+      "loss": 0.3636,
+      "step": 15410
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0003261477511069681,
+      "loss": 0.3286,
+      "step": 15420
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0003260312281519459,
+      "loss": 0.3619,
+      "step": 15430
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0003259147051969238,
+      "loss": 0.4448,
+      "step": 15440
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00032579818224190166,
+      "loss": 0.4407,
+      "step": 15450
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0003256816592868795,
+      "loss": 0.3583,
+      "step": 15460
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00032556513633185736,
+      "loss": 0.4122,
+      "step": 15470
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003254486133768352,
+      "loss": 0.4129,
+      "step": 15480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003253320904218131,
+      "loss": 0.3352,
+      "step": 15490
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00032521556746679096,
+      "loss": 0.4393,
+      "step": 15500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003250990445117688,
+      "loss": 0.3783,
+      "step": 15510
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00032498252155674666,
+      "loss": 0.4093,
+      "step": 15520
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00032486599860172456,
+      "loss": 0.4444,
+      "step": 15530
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003247494756467024,
+      "loss": 0.3857,
+      "step": 15540
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00032463295269168026,
+      "loss": 0.4728,
+      "step": 15550
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003245164297366581,
+      "loss": 0.3692,
+      "step": 15560
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000324399906781636,
+      "loss": 0.3148,
+      "step": 15570
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00032428338382661386,
+      "loss": 0.3775,
+      "step": 15580
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003241668608715917,
+      "loss": 0.4214,
+      "step": 15590
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00032405033791656955,
+      "loss": 0.3387,
+      "step": 15600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00032393381496154745,
+      "loss": 0.4134,
+      "step": 15610
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00032381729200652525,
+      "loss": 0.3061,
+      "step": 15620
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00032370076905150315,
+      "loss": 0.4205,
+      "step": 15630
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000323584246096481,
+      "loss": 0.4003,
+      "step": 15640
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003234677231414589,
+      "loss": 0.4012,
+      "step": 15650
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003233512001864367,
+      "loss": 0.4849,
+      "step": 15660
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003232346772314146,
+      "loss": 0.3454,
+      "step": 15670
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00032311815427639245,
+      "loss": 0.3914,
+      "step": 15680
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00032300163132137035,
+      "loss": 0.4868,
+      "step": 15690
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00032288510836634814,
+      "loss": 0.4,
+      "step": 15700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00032276858541132605,
+      "loss": 0.4648,
+      "step": 15710
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003226520624563039,
+      "loss": 0.3908,
+      "step": 15720
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003225355395012818,
+      "loss": 0.51,
+      "step": 15730
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003224190165462596,
+      "loss": 0.3576,
+      "step": 15740
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003223024935912375,
+      "loss": 0.3894,
+      "step": 15750
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00032218597063621534,
+      "loss": 0.4271,
+      "step": 15760
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00032206944768119325,
+      "loss": 0.389,
+      "step": 15770
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00032195292472617104,
+      "loss": 0.4048,
+      "step": 15780
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00032183640177114894,
+      "loss": 0.3575,
+      "step": 15790
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003217198788161268,
+      "loss": 0.4004,
+      "step": 15800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00032160335586110464,
+      "loss": 0.3556,
+      "step": 15810
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003214868329060825,
+      "loss": 0.3672,
+      "step": 15820
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003213703099510604,
+      "loss": 0.3894,
+      "step": 15830
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00032125378699603824,
+      "loss": 0.3964,
+      "step": 15840
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003211372640410161,
+      "loss": 0.4496,
+      "step": 15850
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00032102074108599394,
+      "loss": 0.3854,
+      "step": 15860
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00032090421813097184,
+      "loss": 0.428,
+      "step": 15870
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003207876951759497,
+      "loss": 0.3453,
+      "step": 15880
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00032067117222092753,
+      "loss": 0.3847,
+      "step": 15890
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003205546492659054,
+      "loss": 0.3499,
+      "step": 15900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003204381263108833,
+      "loss": 0.3786,
+      "step": 15910
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00032032160335586113,
+      "loss": 0.4293,
+      "step": 15920
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00032020508040083893,
+      "loss": 0.3711,
+      "step": 15930
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00032008855744581683,
+      "loss": 0.3793,
+      "step": 15940
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003199720344907947,
+      "loss": 0.4998,
+      "step": 15950
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003198555115357726,
+      "loss": 0.402,
+      "step": 15960
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003197389885807504,
+      "loss": 0.3963,
+      "step": 15970
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003196224656257283,
+      "loss": 0.336,
+      "step": 15980
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00031950594267070613,
+      "loss": 0.3906,
+      "step": 15990
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000319389419715684,
+      "loss": 0.4359,
+      "step": 16000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003192728967606618,
+      "loss": 0.4425,
+      "step": 16010
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003191563738056397,
+      "loss": 0.3933,
+      "step": 16020
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003190398508506176,
+      "loss": 0.3802,
+      "step": 16030
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003189233278955954,
+      "loss": 0.4211,
+      "step": 16040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00031880680494057327,
+      "loss": 0.4331,
+      "step": 16050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003186902819855512,
+      "loss": 0.4106,
+      "step": 16060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000318573759030529,
+      "loss": 0.326,
+      "step": 16070
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00031845723607550687,
+      "loss": 0.444,
+      "step": 16080
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003183407131204847,
+      "loss": 0.3604,
+      "step": 16090
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003182241901654626,
+      "loss": 0.3256,
+      "step": 16100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00031810766721044047,
+      "loss": 0.4363,
+      "step": 16110
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003179911442554183,
+      "loss": 0.4663,
+      "step": 16120
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00031787462130039617,
+      "loss": 0.4167,
+      "step": 16130
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00031775809834537407,
+      "loss": 0.371,
+      "step": 16140
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003176415753903519,
+      "loss": 0.4051,
+      "step": 16150
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00031752505243532977,
+      "loss": 0.4683,
+      "step": 16160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003174085294803076,
+      "loss": 0.4194,
+      "step": 16170
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003172920065252855,
+      "loss": 0.3889,
+      "step": 16180
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003171754835702633,
+      "loss": 0.3615,
+      "step": 16190
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003170589606152412,
+      "loss": 0.4925,
+      "step": 16200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031694243766021906,
+      "loss": 0.3992,
+      "step": 16210
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031682591470519697,
+      "loss": 0.459,
+      "step": 16220
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031670939175017476,
+      "loss": 0.3699,
+      "step": 16230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031659286879515266,
+      "loss": 0.4306,
+      "step": 16240
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003164763458401305,
+      "loss": 0.3863,
+      "step": 16250
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003163598228851084,
+      "loss": 0.4378,
+      "step": 16260
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003162432999300862,
+      "loss": 0.486,
+      "step": 16270
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003161267769750641,
+      "loss": 0.3251,
+      "step": 16280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031601025402004196,
+      "loss": 0.3267,
+      "step": 16290
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031589373106501986,
+      "loss": 0.3773,
+      "step": 16300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00031577720810999765,
+      "loss": 0.3589,
+      "step": 16310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00031566068515497556,
+      "loss": 0.4671,
+      "step": 16320
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003155441621999534,
+      "loss": 0.4616,
+      "step": 16330
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00031542763924493125,
+      "loss": 0.3728,
+      "step": 16340
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003153111162899091,
+      "loss": 0.4246,
+      "step": 16350
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000315194593334887,
+      "loss": 0.4186,
+      "step": 16360
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00031507807037986485,
+      "loss": 0.4021,
+      "step": 16370
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003149615474248427,
+      "loss": 0.3895,
+      "step": 16380
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00031484502446982055,
+      "loss": 0.4319,
+      "step": 16390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00031472850151479845,
+      "loss": 0.3498,
+      "step": 16400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0003146119785597763,
+      "loss": 0.3789,
+      "step": 16410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0003144954556047541,
+      "loss": 0.3468,
+      "step": 16420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.000314378932649732,
+      "loss": 0.4274,
+      "step": 16430
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031426240969470985,
+      "loss": 0.4027,
+      "step": 16440
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031414588673968775,
+      "loss": 0.444,
+      "step": 16450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031402936378466554,
+      "loss": 0.41,
+      "step": 16460
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00031391284082964345,
+      "loss": 0.4299,
+      "step": 16470
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003137963178746213,
+      "loss": 0.4295,
+      "step": 16480
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003136797949195992,
+      "loss": 0.4359,
+      "step": 16490
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000313563271964577,
+      "loss": 0.4861,
+      "step": 16500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003134467490095549,
+      "loss": 0.4234,
+      "step": 16510
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00031333022605453274,
+      "loss": 0.4391,
+      "step": 16520
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0003132137030995106,
+      "loss": 0.3835,
+      "step": 16530
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00031309718014448844,
+      "loss": 0.4576,
+      "step": 16540
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00031298065718946634,
+      "loss": 0.3758,
+      "step": 16550
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0003128641342344442,
+      "loss": 0.3515,
+      "step": 16560
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00031274761127942204,
+      "loss": 0.4279,
+      "step": 16570
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0003126310883243999,
+      "loss": 0.3725,
+      "step": 16580
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0003125145653693778,
+      "loss": 0.4366,
+      "step": 16590
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00031239804241435564,
+      "loss": 0.3677,
+      "step": 16600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0003122815194593335,
+      "loss": 0.4184,
+      "step": 16610
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00031216499650431133,
+      "loss": 0.3896,
+      "step": 16620
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00031204847354928924,
+      "loss": 0.4163,
+      "step": 16630
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0003119319505942671,
+      "loss": 0.3889,
+      "step": 16640
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00031181542763924493,
+      "loss": 0.455,
+      "step": 16650
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0003116989046842228,
+      "loss": 0.3758,
+      "step": 16660
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0003115823817292007,
+      "loss": 0.3656,
+      "step": 16670
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00031146585877417853,
+      "loss": 0.3781,
+      "step": 16680
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0003113493358191564,
+      "loss": 0.3993,
+      "step": 16690
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00031123281286413423,
+      "loss": 0.4871,
+      "step": 16700
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00031111628990911213,
+      "loss": 0.4029,
+      "step": 16710
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0003109997669540899,
+      "loss": 0.3955,
+      "step": 16720
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00031088324399906783,
+      "loss": 0.4014,
+      "step": 16730
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003107667210440457,
+      "loss": 0.4119,
+      "step": 16740
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003106501980890236,
+      "loss": 0.4613,
+      "step": 16750
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003105336751340014,
+      "loss": 0.3741,
+      "step": 16760
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003104171521789793,
+      "loss": 0.4436,
+      "step": 16770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0003103006292239571,
+      "loss": 0.4412,
+      "step": 16780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00031018410626893503,
+      "loss": 0.4534,
+      "step": 16790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0003100675833139128,
+      "loss": 0.4121,
+      "step": 16800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0003099510603588907,
+      "loss": 0.4208,
+      "step": 16810
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00030983453740386857,
+      "loss": 0.4304,
+      "step": 16820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0003097180144488465,
+      "loss": 0.3814,
+      "step": 16830
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00030960149149382427,
+      "loss": 0.4537,
+      "step": 16840
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00030948496853880217,
+      "loss": 0.3598,
+      "step": 16850
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00030936844558378,
+      "loss": 0.4689,
+      "step": 16860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003092519226287579,
+      "loss": 0.4239,
+      "step": 16870
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003091353996737357,
+      "loss": 0.4838,
+      "step": 16880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003090188767187136,
+      "loss": 0.3399,
+      "step": 16890
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00030890235376369147,
+      "loss": 0.4228,
+      "step": 16900
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00030878583080866926,
+      "loss": 0.408,
+      "step": 16910
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00030866930785364717,
+      "loss": 0.3553,
+      "step": 16920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000308552784898625,
+      "loss": 0.4471,
+      "step": 16930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003084362619436029,
+      "loss": 0.3707,
+      "step": 16940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003083197389885807,
+      "loss": 0.4226,
+      "step": 16950
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0003082032160335586,
+      "loss": 0.3847,
+      "step": 16960
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00030808669307853646,
+      "loss": 0.3488,
+      "step": 16970
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00030797017012351436,
+      "loss": 0.3581,
+      "step": 16980
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00030785364716849216,
+      "loss": 0.4423,
+      "step": 16990
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00030773712421347006,
+      "loss": 0.3793,
+      "step": 17000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0003076206012584479,
+      "loss": 0.4583,
+      "step": 17010
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0003075040783034258,
+      "loss": 0.4894,
+      "step": 17020
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0003073875553484036,
+      "loss": 0.4853,
+      "step": 17030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0003072710323933815,
+      "loss": 0.3569,
+      "step": 17040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030715450943835936,
+      "loss": 0.4572,
+      "step": 17050
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030703798648333726,
+      "loss": 0.444,
+      "step": 17060
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030692146352831505,
+      "loss": 0.4176,
+      "step": 17070
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00030680494057329296,
+      "loss": 0.4235,
+      "step": 17080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0003066884176182708,
+      "loss": 0.4773,
+      "step": 17090
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00030657189466324865,
+      "loss": 0.416,
+      "step": 17100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0003064553717082265,
+      "loss": 0.4587,
+      "step": 17110
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0003063388487532044,
+      "loss": 0.4066,
+      "step": 17120
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00030622232579818225,
+      "loss": 0.3455,
+      "step": 17130
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0003061058028431601,
+      "loss": 0.462,
+      "step": 17140
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00030598927988813795,
+      "loss": 0.4757,
+      "step": 17150
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00030587275693311585,
+      "loss": 0.3779,
+      "step": 17160
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0003057562339780937,
+      "loss": 0.3778,
+      "step": 17170
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00030563971102307155,
+      "loss": 0.4172,
+      "step": 17180
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0003055231880680494,
+      "loss": 0.5275,
+      "step": 17190
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0003054066651130273,
+      "loss": 0.4051,
+      "step": 17200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00030529014215800515,
+      "loss": 0.3594,
+      "step": 17210
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.000305173619202983,
+      "loss": 0.4589,
+      "step": 17220
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00030505709624796084,
+      "loss": 0.4427,
+      "step": 17230
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00030494057329293875,
+      "loss": 0.4177,
+      "step": 17240
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0003048240503379166,
+      "loss": 0.3853,
+      "step": 17250
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00030470752738289444,
+      "loss": 0.4029,
+      "step": 17260
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0003045910044278723,
+      "loss": 0.4892,
+      "step": 17270
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0003044744814728502,
+      "loss": 0.3959,
+      "step": 17280
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000304357958517828,
+      "loss": 0.4701,
+      "step": 17290
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003042414355628059,
+      "loss": 0.366,
+      "step": 17300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00030412491260778374,
+      "loss": 0.3826,
+      "step": 17310
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00030400838965276164,
+      "loss": 0.378,
+      "step": 17320
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00030389186669773944,
+      "loss": 0.4033,
+      "step": 17330
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00030377534374271734,
+      "loss": 0.4646,
+      "step": 17340
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0003036588207876952,
+      "loss": 0.4589,
+      "step": 17350
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0003035422978326731,
+      "loss": 0.5188,
+      "step": 17360
+    },
+    {
+      "epoch": 4.0,
+      "eval_Bleu_1": 0.005368360936607346,
+      "eval_Bleu_2": 2.1529258556147605e-11,
+      "eval_Bleu_3": 3.921163253647881e-14,
+      "eval_Bleu_4": 1.83228507309933e-15,
+      "eval_ROUGE_L": 0.008549703525608982,
+      "eval_cer": 0.4694332130837954,
+      "eval_em": 0.5671390013495277,
+      "eval_f1": 0.7247699680587338,
+      "eval_loss": 0.8551194071769714,
+      "eval_runtime": 321.1768,
+      "eval_samples_per_second": 18.457,
+      "eval_steps_per_second": 3.693,
+      "eval_wer": 0.5537877989674428,
+      "step": 17365
+    }
+  ],
+  "max_steps": 43410,
+  "num_train_epochs": 10,
+  "total_flos": 1.3589112280001126e+17,
+  "trial_name": null,
+  "trial_params": null
+}