diff --git "a/checkpoint-8682/trainer_state.json" "b/checkpoint-8682/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-8682/trainer_state.json"
@@ -0,0 +1,5258 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.999884825799021,
+  "global_step": 8682,
+  "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
+    }
+  ],
+  "max_steps": 43410,
+  "num_train_epochs": 10,
+  "total_flos": 6.797235869591962e+16,
+  "trial_name": null,
+  "trial_params": null
+}