diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.8521948508983144,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0009999978838190456,
+      "loss": 2.9794,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0009999915352940948,
+      "loss": 2.3885,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000999980954478887,
+      "loss": 2.3057,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000999966141462985,
+      "loss": 2.2692,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000999947096371777,
+      "loss": 2.2576,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009999238193664748,
+      "loss": 2.2388,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009998963106441117,
+      "loss": 2.2523,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009998645704375414,
+      "loss": 2.218,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999828599015436,
+      "loss": 2.2457,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009997883966822835,
+      "loss": 2.198,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009997439637783859,
+      "loss": 2.2013,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999695300679855,
+      "loss": 2.1765,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009996424077986109,
+      "loss": 2.1741,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999585285582377,
+      "loss": 2.1898,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009995239345146772,
+      "loss": 2.1466,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009994583551148314,
+      "loss": 2.1423,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009993885479379506,
+      "loss": 2.1451,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999314513574934,
+      "loss": 2.202,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009992362526524616,
+      "loss": 2.1208,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009991537658329906,
+      "loss": 2.1591,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999067053814749,
+      "loss": 2.1788,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009989761173317304,
+      "loss": 2.147,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000998880957153687,
+      "loss": 2.1249,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000998781574086123,
+      "loss": 2.165,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998677968970289,
+      "loss": 2.1428,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985701426831735,
+      "loss": 2.1384,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984580961374964,
+      "loss": 2.1585,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983418302817008,
+      "loss": 2.1156,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009982213460999448,
+      "loss": 2.0811,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998096644612094,
+      "loss": 2.1081,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979677268737118,
+      "loss": 2.1246,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978345939760515,
+      "loss": 2.1229,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997697247046046,
+      "loss": 2.1033,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975556872462994,
+      "loss": 2.0931,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997409915775076,
+      "loss": 2.1206,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009972599338662915,
+      "loss": 2.0537,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009971057427895012,
+      "loss": 2.0762,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009969473438498897,
+      "loss": 2.0883,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967847383882604,
+      "loss": 2.1174,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966179277810239,
+      "loss": 2.1111,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009964469134401855,
+      "loss": 2.1288,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009962716968133346,
+      "loss": 2.0967,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009960922793836318,
+      "loss": 2.1216,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009959086626697955,
+      "loss": 2.0924,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009957208482260908,
+      "loss": 2.0809,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009955288376423152,
+      "loss": 2.1082,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009953326325437852,
+      "loss": 2.0885,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009951322345913224,
+      "loss": 2.1133,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009949276454812408,
+      "loss": 2.0844,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099471886694533,
+      "loss": 2.0796,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009945059007508434,
+      "loss": 2.1255,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009942887487004804,
+      "loss": 2.0913,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009940674126323733,
+      "loss": 2.1003,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009938418944200709,
+      "loss": 2.0541,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009936121959725223,
+      "loss": 2.0523,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009933783192340618,
+      "loss": 2.1225,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009931402661843911,
+      "loss": 2.0446,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000992898038838564,
+      "loss": 2.0921,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009926516392469674,
+      "loss": 2.1081,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009924010694953064,
+      "loss": 2.0734,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009921463317045843,
+      "loss": 2.0652,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009918874280310862,
+      "loss": 2.0818,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009916243606663605,
+      "loss": 2.0776,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009913571318371994,
+      "loss": 2.1025,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009910857438056215,
+      "loss": 2.066,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009908101988688512,
+      "loss": 2.0575,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009905304993593008,
+      "loss": 2.1269,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009902466476445486,
+      "loss": 2.0518,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009899586461273218,
+      "loss": 2.0698,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000989666497245473,
+      "loss": 2.0988,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009893702034719624,
+      "loss": 2.0986,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009890697673148345,
+      "loss": 2.0237,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009887651913171986,
+      "loss": 2.0027,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009884564780572064,
+      "loss": 2.0563,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009881436301480305,
+      "loss": 2.0624,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000987826650237842,
+      "loss": 2.0926,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000987505541009788,
+      "loss": 2.0585,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009871803051819696,
+      "loss": 2.0494,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009868509455074183,
+      "loss": 2.0106,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009865174647740729,
+      "loss": 2.0861,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009861798658047556,
+      "loss": 2.0478,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009858381514571484,
+      "loss": 2.0469,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000985492324623769,
+      "loss": 2.0671,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009851423882319458,
+      "loss": 2.0808,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009847883452437937,
+      "loss": 2.0331,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009844301986561893,
+      "loss": 2.0295,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000984067951500744,
+      "loss": 2.0873,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00098370160684378,
+      "loss": 2.1038,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009833311677863042,
+      "loss": 2.0337,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009829566374639801,
+      "loss": 2.0407,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009825780190471042,
+      "loss": 2.1049,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000982195315740576,
+      "loss": 2.0475,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009818085307838741,
+      "loss": 2.0624,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000981417667451026,
+      "loss": 2.0714,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009810227290505816,
+      "loss": 2.0947,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009806237189255859,
+      "loss": 2.0591,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009802206404535489,
+      "loss": 2.0301,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000979813497046419,
+      "loss": 2.0556,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009794022921505523,
+      "loss": 2.0753,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000978987029246685,
+      "loss": 2.0898,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009785677118499029,
+      "loss": 2.0464,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009781443435096116,
+      "loss": 2.0828,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009777169278095074,
+      "loss": 2.1137,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009772854683675462,
+      "loss": 2.0167,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000976849968835913,
+      "loss": 2.07,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009764104329009909,
+      "loss": 2.0409,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009759668642833304,
+      "loss": 2.015,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009755192667376173,
+      "loss": 2.0175,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009750676440526411,
+      "loss": 2.0773,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009746120000512632,
+      "loss": 2.0245,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009741523385903841,
+      "loss": 2.094,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009736886635609112,
+      "loss": 2.0506,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009732209788877258,
+      "loss": 2.0287,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009727492885296489,
+      "loss": 2.1162,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009722735964794099,
+      "loss": 2.1096,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009717939067636099,
+      "loss": 2.0621,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009713102234426903,
+      "loss": 2.0796,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009708225506108965,
+      "loss": 2.0565,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009703308923962447,
+      "loss": 2.0669,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009698352529604857,
+      "loss": 2.0638,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009693356364990705,
+      "loss": 2.0358,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009688320472411143,
+      "loss": 2.0859,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009683244894493613,
+      "loss": 2.0932,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009678129674201479,
+      "loss": 2.0129,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009672974854833669,
+      "loss": 2.055,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009667780480024304,
+      "loss": 2.0665,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009662546593742334,
+      "loss": 2.0488,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009657273240291159,
+      "loss": 2.0543,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009651960464308261,
+      "loss": 2.0418,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009646608310764819,
+      "loss": 2.033,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009641216824965338,
+      "loss": 2.1034,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009635786052547253,
+      "loss": 2.0866,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009630316039480556,
+      "loss": 2.0607,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009624806832067394,
+      "loss": 2.0457,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009619258476941686,
+      "loss": 2.032,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000961367102106873,
+      "loss": 2.0519,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009608044511744791,
+      "loss": 2.0449,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009602378996596721,
+      "loss": 1.9949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009596674523581539,
+      "loss": 2.0394,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009590931140986035,
+      "loss": 2.0386,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009585148897426354,
+      "loss": 2.0254,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009579327841847593,
+      "loss": 2.0238,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000957346802352338,
+      "loss": 2.0509,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009567569492055456,
+      "loss": 2.0004,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009561632297373263,
+      "loss": 2.0203,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009555656489733513,
+      "loss": 2.0182,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000954964211971977,
+      "loss": 1.9754,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009543589238242012,
+      "loss": 2.0374,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000953749789653621,
+      "loss": 2.0367,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000953136814616389,
+      "loss": 2.0866,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009525200039011694,
+      "loss": 2.0083,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009518993627290948,
+      "loss": 2.0525,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009512748963537212,
+      "loss": 2.0636,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000950646610060984,
+      "loss": 2.0522,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009500145091691532,
+      "loss": 2.05,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009493785990287882,
+      "loss": 1.9887,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009487388850226926,
+      "loss": 2.0309,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000948095372565869,
+      "loss": 1.9954,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009474480671054726,
+      "loss": 2.0078,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009467969741207652,
+      "loss": 2.0395,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009461420991230693,
+      "loss": 2.0415,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009454834476557207,
+      "loss": 2.0308,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009448210252940223,
+      "loss": 2.0826,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009441548376451963,
+      "loss": 2.0424,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009434848903483373,
+      "loss": 2.0125,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009428111890743639,
+      "loss": 2.0139,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009421337395259717,
+      "loss": 2.0682,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009414525474375837,
+      "loss": 2.0577,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009407676185753029,
+      "loss": 2.0262,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009400789587368632,
+      "loss": 2.0515,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009393865737515794,
+      "loss": 2.0398,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009386904694802997,
+      "loss": 2.0146,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009379906518153543,
+      "loss": 2.0438,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009372871266805063,
+      "loss": 2.0377,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000936579900030902,
+      "loss": 2.0789,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009358689778530193,
+      "loss": 2.0201,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009351543661646185,
+      "loss": 2.0114,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009344360710146898,
+      "loss": 2.0242,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009337140984834034,
+      "loss": 2.0436,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009329884546820572,
+      "loss": 2.0452,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000932259145753026,
+      "loss": 2.0254,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009315261778697083,
+      "loss": 2.0409,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009307895572364746,
+      "loss": 2.0301,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009300492900886154,
+      "loss": 2.0078,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009293053826922873,
+      "loss": 1.9851,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009285578413444613,
+      "loss": 1.9947,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009278066723728682,
+      "loss": 2.0331,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009270518821359461,
+      "loss": 2.0058,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009262934770227858,
+      "loss": 2.05,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009255314634530771,
+      "loss": 2.0444,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009247658478770543,
+      "loss": 2.0045,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000923996636775442,
+      "loss": 2.0211,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009232238366593997,
+      "loss": 2.0124,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009224474540704671,
+      "loss": 2.0067,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009216674955805079,
+      "loss": 2.0247,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009208839677916557,
+      "loss": 2.0314,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009200968773362568,
+      "loss": 2.067,
+      "step": 1970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009193062308768145,
+      "loss": 2.0168,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009185120351059326,
+      "loss": 2.0649,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009177142967462591,
+      "loss": 2.0208,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009169130225504289,
+      "loss": 2.016,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009161082193010066,
+      "loss": 2.0331,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009152998938104296,
+      "loss": 2.0757,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009144880529209498,
+      "loss": 2.0093,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009136727035045765,
+      "loss": 1.954,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009128538524630167,
+      "loss": 2.0062,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009120315067276187,
+      "loss": 2.0179,
+      "step": 2070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009112056732593119,
+      "loss": 2.0301,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009103763590485488,
+      "loss": 2.0223,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009095435711152449,
+      "loss": 2.0208,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00090870731650872,
+      "loss": 2.0492,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009078676023076385,
+      "loss": 1.9917,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009070244356199492,
+      "loss": 2.0235,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009061778235828253,
+      "loss": 2.01,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009054129328980229,
+      "loss": 1.9999,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000904559794464287,
+      "loss": 2.0022,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009037032315436229,
+      "loss": 2.0004,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009028432513865994,
+      "loss": 1.9936,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000901979861272711,
+      "loss": 2.0098,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009011130685103166,
+      "loss": 1.9829,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009002428804365775,
+      "loss": 2.0135,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0008993693044173954,
+      "loss": 2.0526,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0008984923478473499,
+      "loss": 2.0145,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0008976120181496365,
+      "loss": 2.0364,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0008967283227760028,
+      "loss": 2.019,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000895841269206686,
+      "loss": 1.9874,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0008949508649503495,
+      "loss": 2.0699,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0008940571175440196,
+      "loss": 2.0407,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0008931600345530213,
+      "loss": 2.0176,
+      "step": 2290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0008922596235709142,
+      "loss": 2.0235,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0008913558922194287,
+      "loss": 2.0268,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000890448848148401,
+      "loss": 2.0279,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0008895384990357087,
+      "loss": 2.0073,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0008886248525872056,
+      "loss": 2.0262,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0008877079165366566,
+      "loss": 2.0112,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000886787698645672,
+      "loss": 2.042,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0008858642067036424,
+      "loss": 2.0129,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0008849374485276715,
+      "loss": 2.0264,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000884007431962512,
+      "loss": 2.0211,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0008830741648804967,
+      "loss": 2.0175,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0008821376551814739,
+      "loss": 2.0423,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0008811979107927396,
+      "loss": 2.0112,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0008802549396689705,
+      "loss": 2.0422,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0008793087497921566,
+      "loss": 2.0075,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0008784544334882426,
+      "loss": 1.9958,
+      "step": 2450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000877502150068553,
+      "loss": 2.0348,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008765466711974225,
+      "loss": 2.0232,
+      "step": 2470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008755880049627157,
+      "loss": 1.9953,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008746261594792776,
+      "loss": 2.0223,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008736611428888644,
+      "loss": 2.0332,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008726929633600753,
+      "loss": 2.0075,
+      "step": 2510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008717216290882824,
+      "loss": 2.002,
+      "step": 2520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008707471482955622,
+      "loss": 2.0594,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008697695292306253,
+      "loss": 2.0214,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008687887801687473,
+      "loss": 1.994,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008678049094116979,
+      "loss": 2.0338,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008668179252876714,
+      "loss": 2.0176,
+      "step": 2570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008658278361512161,
+      "loss": 2.0013,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008648346503831627,
+      "loss": 2.0141,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008638383763905546,
+      "loss": 1.9655,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008628390226065761,
+      "loss": 2.039,
+      "step": 2610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008618365974904808,
+      "loss": 1.9418,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008608311095275205,
+      "loss": 2.011,
+      "step": 2630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008598225672288731,
+      "loss": 2.0174,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008588109791315707,
+      "loss": 1.984,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000857796353798427,
+      "loss": 2.0437,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008567786998179654,
+      "loss": 2.0255,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008557580258043459,
+      "loss": 2.0067,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008547343403972918,
+      "loss": 2.0201,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008537076522620175,
+      "loss": 2.0258,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008526779700891545,
+      "loss": 1.9977,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008516453025946781,
+      "loss": 1.9835,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008506096585198333,
+      "loss": 2.0428,
+      "step": 2730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008495710466310613,
+      "loss": 1.9554,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008485294757199248,
+      "loss": 2.0497,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008474849546030339,
+      "loss": 2.001,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008464374921219714,
+      "loss": 1.9824,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008453870971432179,
+      "loss": 2.0378,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008443337785580769,
+      "loss": 1.9756,
+      "step": 2790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008432775452825992,
+      "loss": 2.0563,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008422184062575077,
+      "loss": 2.0115,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008411563704481217,
+      "loss": 2.0301,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008400914468442814,
+      "loss": 1.9912,
+      "step": 2830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008390236444602706,
+      "loss": 2.0669,
+      "step": 2840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008379529723347417,
+      "loss": 2.0138,
+      "step": 2850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008368794395306386,
+      "loss": 2.0486,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008358030551351199,
+      "loss": 2.0342,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008347238282594823,
+      "loss": 2.0474,
+      "step": 2880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008336417680390829,
+      "loss": 2.006,
+      "step": 2890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008325568836332633,
+      "loss": 1.99,
+      "step": 2900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00083146918422527,
+      "loss": 2.0148,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008303786790221778,
+      "loss": 2.0381,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008292853772548126,
+      "loss": 2.0591,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008281892881776714,
+      "loss": 2.0238,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008270904210688459,
+      "loss": 1.9939,
+      "step": 2950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008259887852299427,
+      "loss": 1.9941,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008248843899860045,
+      "loss": 2.0193,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008237772446854325,
+      "loss": 2.0166,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008226673586999058,
+      "loss": 2.0219,
+      "step": 2990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008215547414243025,
+      "loss": 2.0078,
+      "step": 3000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008204394022766208,
+      "loss": 2.0022,
+      "step": 3010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008193213506978983,
+      "loss": 2.0009,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008182005961521331,
+      "loss": 2.0197,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008170771481262027,
+      "loss": 2.0101,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008159510161297846,
+      "loss": 2.0307,
+      "step": 3050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008148222096952748,
+      "loss": 2.0075,
+      "step": 3060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008136907383777081,
+      "loss": 2.0415,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000812556611754677,
+      "loss": 2.0031,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008114198394262497,
+      "loss": 2.0025,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008102804310148905,
+      "loss": 1.983,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008091383961653765,
+      "loss": 2.0056,
+      "step": 3110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008079937445447175,
+      "loss": 1.9679,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008068464858420732,
+      "loss": 1.9897,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008056966297686717,
+      "loss": 1.9833,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008045441860577273,
+      "loss": 2.0088,
+      "step": 3150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008033891644643576,
+      "loss": 1.9985,
+      "step": 3160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008022315747655011,
+      "loss": 2.013,
+      "step": 3170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008010714267598354,
+      "loss": 1.9902,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007999087302676926,
+      "loss": 2.0133,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007987434951309774,
+      "loss": 2.0092,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007975757312130836,
+      "loss": 1.9894,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007964054483988103,
+      "loss": 1.9742,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000795232656594278,
+      "loss": 2.016,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007940573657268458,
+      "loss": 2.0386,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007928795857450259,
+      "loss": 1.9602,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007916993266184007,
+      "loss": 1.9681,
+      "step": 3260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007905165983375379,
+      "loss": 2.0326,
+      "step": 3270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007893314109139058,
+      "loss": 1.9958,
+      "step": 3280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007881437743797882,
+      "loss": 2.0077,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007869536987882008,
+      "loss": 1.9813,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007857611942128044,
+      "loss": 1.9538,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007845662707478214,
+      "loss": 1.9919,
+      "step": 3320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007833689385079485,
+      "loss": 1.9922,
+      "step": 3330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007821692076282725,
+      "loss": 1.9973,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007809670882641842,
+      "loss": 2.0231,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007797625905912915,
+      "loss": 1.9971,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007785557248053351,
+      "loss": 1.9934,
+      "step": 3370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007773465011221002,
+      "loss": 1.9855,
+      "step": 3380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007761349297773315,
+      "loss": 2.0421,
+      "step": 3390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007749210210266457,
+      "loss": 1.9795,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007737047851454451,
+      "loss": 2.0006,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007724862324288309,
+      "loss": 2.011,
+      "step": 3420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007712653731915147,
+      "loss": 1.9969,
+      "step": 3430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007700422177677331,
+      "loss": 1.9581,
+      "step": 3440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007688167765111592,
+      "loss": 1.9858,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007675890597948141,
+      "loss": 1.9945,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007663590780109814,
+      "loss": 2.0116,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007651268415711168,
+      "loss": 1.9831,
+      "step": 3480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007638923609057616,
+      "loss": 2.0113,
+      "step": 3490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007626556464644538,
+      "loss": 1.9927,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007614167087156393,
+      "loss": 2.0209,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007601755581465844,
+      "loss": 1.9941,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007589322052632855,
+      "loss": 2.0443,
+      "step": 3530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007576866605903812,
+      "loss": 1.975,
+      "step": 3540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007564389346710638,
+      "loss": 1.9655,
+      "step": 3550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007551890380669879,
+      "loss": 2.0034,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007539369813581836,
+      "loss": 1.953,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000752682775142965,
+      "loss": 2.0162,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007514264300378411,
+      "loss": 2.0096,
+      "step": 3590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007501679566774267,
+      "loss": 2.0431,
+      "step": 3600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007489073657143506,
+      "loss": 1.961,
+      "step": 3610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007476446678191676,
+      "loss": 2.0179,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007463798736802665,
+      "loss": 1.9885,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007451129940037804,
+      "loss": 1.9955,
+      "step": 3640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000743844039513496,
+      "loss": 1.9671,
+      "step": 3650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007425730209507622,
+      "loss": 1.9866,
+      "step": 3660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007412999490744008,
+      "loss": 1.9508,
+      "step": 3670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007400248346606129,
+      "loss": 2.0261,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007387476885028902,
+      "loss": 1.9946,
+      "step": 3690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007374685214119221,
+      "loss": 1.969,
+      "step": 3700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007361873442155049,
+      "loss": 1.9802,
+      "step": 3710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007349041677584496,
+      "loss": 1.9739,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007336190029024907,
+      "loss": 1.964,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007323318605261933,
+      "loss": 2.0009,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007310427515248626,
+      "loss": 1.9825,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007297516868104501,
+      "loss": 1.9744,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000728458677311462,
+      "loss": 2.0197,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007271637339728667,
+      "loss": 2.0016,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007258668677560019,
+      "loss": 1.9721,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007245680896384818,
+      "loss": 1.9591,
+      "step": 3800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007232674106141048,
+      "loss": 1.9762,
+      "step": 3810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007219648416927592,
+      "loss": 2.0026,
+      "step": 3820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007206603939003314,
+      "loss": 1.9876,
+      "step": 3830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007193540782786117,
+      "loss": 1.9991,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007180459058852011,
+      "loss": 1.9897,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007167358877934176,
+      "loss": 2.0581,
+      "step": 3860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007154240350922025,
+      "loss": 1.9588,
+      "step": 3870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007141103588860265,
+      "loss": 2.0151,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000712794870294796,
+      "loss": 1.9494,
+      "step": 3890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007114775804537586,
+      "loss": 1.9936,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007101585005134088,
+      "loss": 2.0016,
+      "step": 3910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007088376416393944,
+      "loss": 1.9628,
+      "step": 3920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007075150150124208,
+      "loss": 1.9955,
+      "step": 3930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007061906318281568,
+      "loss": 1.9731,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007048645032971407,
+      "loss": 1.9554,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007035366406446839,
+      "loss": 1.985,
+      "step": 3960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007022070551107774,
+      "loss": 1.9961,
+      "step": 3970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007008757579499954,
+      "loss": 2.0112,
+      "step": 3980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0006995427604314004,
+      "loss": 2.0086,
+      "step": 3990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0006982080738384487,
+      "loss": 2.019,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0006968717094688933,
+      "loss": 1.9983,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0006955336786346898,
+      "loss": 1.981,
+      "step": 4020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0006941939926618997,
+      "loss": 1.9878,
+      "step": 4030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0006928526628905942,
+      "loss": 2.0067,
+      "step": 4040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0006915097006747601,
+      "loss": 2.035,
+      "step": 4050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0006901651173822013,
+      "loss": 1.9614,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0006888189243944439,
+      "loss": 1.9655,
+      "step": 4070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0006874711331066401,
+      "loss": 1.9692,
+      "step": 4080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0006861217549274705,
+      "loss": 1.9791,
+      "step": 4090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0006847708012790492,
+      "loss": 1.9804,
+      "step": 4100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0006834182835968253,
+      "loss": 1.9973,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0006820642133294876,
+      "loss": 2.0153,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0006807086019388669,
+      "loss": 1.9639,
+      "step": 4130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0006793514608998393,
+      "loss": 2.0162,
+      "step": 4140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000677992801700229,
+      "loss": 1.994,
+      "step": 4150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0006766326358407104,
+      "loss": 1.9992,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0006752709748347125,
+      "loss": 1.976,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000673907830208319,
+      "loss": 1.9452,
+      "step": 4180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006725432135001732,
+      "loss": 1.9982,
+      "step": 4190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006711771362613785,
+      "loss": 1.9766,
+      "step": 4200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006698096100554014,
+      "loss": 1.9838,
+      "step": 4210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006684406464579735,
+      "loss": 1.9746,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006670702570569937,
+      "loss": 1.9838,
+      "step": 4230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006656984534524296,
+      "loss": 1.9747,
+      "step": 4240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006643252472562201,
+      "loss": 1.9896,
+      "step": 4250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000662950650092176,
+      "loss": 1.9607,
+      "step": 4260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006615746735958832,
+      "loss": 2.0075,
+      "step": 4270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006603351250463683,
+      "loss": 1.9918,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000658956559916577,
+      "loss": 1.9669,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006575766492634068,
+      "loss": 1.9645,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006561954047674206,
+      "loss": 1.9511,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006548128381204714,
+      "loss": 1.9934,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006534289610256039,
+      "loss": 1.9973,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006520437851969557,
+      "loss": 1.9983,
+      "step": 4340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006506573223596575,
+      "loss": 1.9906,
+      "step": 4350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006492695842497347,
+      "loss": 1.9392,
+      "step": 4360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006478805826140066,
+      "loss": 2.0015,
+      "step": 4370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006464903292099886,
+      "loss": 2.0168,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006450988358057919,
+      "loss": 1.9676,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006437061141800238,
+      "loss": 1.9798,
+      "step": 4400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006423121761216881,
+      "loss": 1.9753,
+      "step": 4410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006409170334300858,
+      "loss": 1.9965,
+      "step": 4420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006395206979147141,
+      "loss": 1.9532,
+      "step": 4430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006381231813951678,
+      "loss": 1.9962,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006367244957010379,
+      "loss": 1.9758,
+      "step": 4450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000635324652671813,
+      "loss": 1.9606,
+      "step": 4460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006339236641567774,
+      "loss": 1.9785,
+      "step": 4470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006325215420149119,
+      "loss": 1.9334,
+      "step": 4480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006311182981147935,
+      "loss": 1.9829,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000629713944334494,
+      "loss": 1.9671,
+      "step": 4500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006283084925614805,
+      "loss": 2.0108,
+      "step": 4510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006269019546925139,
+      "loss": 1.9341,
+      "step": 4520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006254943426335488,
+      "loss": 1.9378,
+      "step": 4530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006240856682996326,
+      "loss": 1.9619,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006226759436148044,
+      "loss": 1.9516,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006212651805119946,
+      "loss": 1.9882,
+      "step": 4560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000619853390932923,
+      "loss": 1.9636,
+      "step": 4570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006184405868279985,
+      "loss": 1.9894,
+      "step": 4580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006170267801562176,
+      "loss": 1.9365,
+      "step": 4590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006156119828850633,
+      "loss": 2.0264,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006141962069904041,
+      "loss": 1.9659,
+      "step": 4610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006127794644563915,
+      "loss": 1.9368,
+      "step": 4620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006113617672753598,
+      "loss": 1.959,
+      "step": 4630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006099431274477245,
+      "loss": 1.9832,
+      "step": 4640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006085235569818795,
+      "loss": 1.9572,
+      "step": 4650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006071030678940969,
+      "loss": 1.99,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006056816722084248,
+      "loss": 1.9963,
+      "step": 4670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006042593819565846,
+      "loss": 1.9997,
+      "step": 4680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006028362091778706,
+      "loss": 1.9221,
+      "step": 4690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006014121659190476,
+      "loss": 1.9623,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005999872642342484,
+      "loss": 1.9949,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005985615161848721,
+      "loss": 1.9688,
+      "step": 4720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000597134933839482,
+      "loss": 1.975,
+      "step": 4730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005957075292737042,
+      "loss": 2.0048,
+      "step": 4740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005942793145701235,
+      "loss": 1.9591,
+      "step": 4750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005928503018181833,
+      "loss": 1.9957,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005914205031140819,
+      "loss": 1.9823,
+      "step": 4770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005899899305606703,
+      "loss": 2.0032,
+      "step": 4780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005885585962673501,
+      "loss": 1.9722,
+      "step": 4790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005871265123499708,
+      "loss": 1.9903,
+      "step": 4800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005856936909307273,
+      "loss": 2.0258,
+      "step": 4810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000584260144138057,
+      "loss": 2.0444,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005828258841065377,
+      "loss": 1.968,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005813909229767846,
+      "loss": 1.9598,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005799552728953472,
+      "loss": 2.0047,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000578518946014607,
+      "loss": 1.939,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005770819544926741,
+      "loss": 1.9835,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005756443104932852,
+      "loss": 2.028,
+      "step": 4880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005742060261856996,
+      "loss": 1.973,
+      "step": 4890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005727671137445967,
+      "loss": 1.9734,
+      "step": 4900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005713275853499729,
+      "loss": 1.9921,
+      "step": 4910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005698874531870386,
+      "loss": 1.9607,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005684467294461146,
+      "loss": 1.9468,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005670054263225297,
+      "loss": 1.9524,
+      "step": 4940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005655635560165166,
+      "loss": 1.9737,
+      "step": 4950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005641211307331093,
+      "loss": 1.9532,
+      "step": 4960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005626781626820395,
+      "loss": 2.0048,
+      "step": 4970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005612346640776331,
+      "loss": 2.0276,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005597906471387074,
+      "loss": 1.9476,
+      "step": 4990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005583461240884666,
+      "loss": 1.943,
+      "step": 5000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005569011071543997,
+      "loss": 1.9859,
+      "step": 5010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005554556085681757,
+      "loss": 1.9607,
+      "step": 5020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005540096405655409,
+      "loss": 1.9825,
+      "step": 5030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005525632153862152,
+      "loss": 1.9875,
+      "step": 5040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005511163452737884,
+      "loss": 2.0051,
+      "step": 5050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005496690424756164,
+      "loss": 2.0089,
+      "step": 5060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005482213192427174,
+      "loss": 1.9495,
+      "step": 5070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005467731878296688,
+      "loss": 1.9701,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005453246604945034,
+      "loss": 1.9588,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005438757494986044,
+      "loss": 1.9833,
+      "step": 5100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005424264671066038,
+      "loss": 1.9974,
+      "step": 5110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005409768255862766,
+      "loss": 1.9746,
+      "step": 5120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005395268372084378,
+      "loss": 1.9757,
+      "step": 5130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005380765142468387,
+      "loss": 1.9908,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005366258689780625,
+      "loss": 2.006,
+      "step": 5150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005351749136814208,
+      "loss": 1.9957,
+      "step": 5160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005337236606388495,
+      "loss": 1.959,
+      "step": 5170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005322721221348047,
+      "loss": 2.0027,
+      "step": 5180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000530820310456159,
+      "loss": 1.9248,
+      "step": 5190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005293682378920974,
+      "loss": 1.9742,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000527915916734013,
+      "loss": 1.9844,
+      "step": 5210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005264633592754035,
+      "loss": 1.9985,
+      "step": 5220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005250105778117661,
+      "loss": 2.0074,
+      "step": 5230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005235575846404952,
+      "loss": 1.9901,
+      "step": 5240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005221043920607763,
+      "loss": 1.9861,
+      "step": 5250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005206510123734837,
+      "loss": 1.9576,
+      "step": 5260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005191974578810746,
+      "loss": 1.9741,
+      "step": 5270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005177437408874868,
+      "loss": 1.9794,
+      "step": 5280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005162898736980327,
+      "loss": 1.9883,
+      "step": 5290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005148358686192967,
+      "loss": 1.969,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005133817379590302,
+      "loss": 2.0018,
+      "step": 5310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005119274940260476,
+      "loss": 1.971,
+      "step": 5320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005104731491301225,
+      "loss": 1.9564,
+      "step": 5330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005090187155818823,
+      "loss": 1.9798,
+      "step": 5340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005075642056927057,
+      "loss": 2.0331,
+      "step": 5350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000506109631774617,
+      "loss": 1.9505,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005046550061401825,
+      "loss": 1.9348,
+      "step": 5370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005032003411024067,
+      "loss": 2.0099,
+      "step": 5380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005017456489746275,
+      "loss": 1.9756,
+      "step": 5390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005002909420704116,
+      "loss": 1.9444,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004988362327034516,
+      "loss": 1.976,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004973815331874603,
+      "loss": 1.9593,
+      "step": 5420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004959268558360675,
+      "loss": 1.9251,
+      "step": 5430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004944722129627151,
+      "loss": 1.972,
+      "step": 5440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004930176168805531,
+      "loss": 1.9307,
+      "step": 5450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004915630799023359,
+      "loss": 1.9541,
+      "step": 5460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004901086143403172,
+      "loss": 1.9212,
+      "step": 5470
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004886542325061464,
+      "loss": 1.965,
+      "step": 5480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004871999467107639,
+      "loss": 2.0083,
+      "step": 5490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004857457692642974,
+      "loss": 1.9676,
+      "step": 5500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000484291712475957,
+      "loss": 1.9384,
+      "step": 5510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00048283778865393243,
+      "loss": 1.9339,
+      "step": 5520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0004813840101052869,
+      "loss": 2.0182,
+      "step": 5530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004799303891358545,
+      "loss": 1.9262,
+      "step": 5540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004784769380501352,
+      "loss": 1.9397,
+      "step": 5550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004770236691511908,
+      "loss": 1.9404,
+      "step": 5560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004755705947405417,
+      "loss": 1.9415,
+      "step": 5570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00047411772711806105,
+      "loss": 1.9292,
+      "step": 5580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000472665078581872,
+      "loss": 1.9487,
+      "step": 5590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00047121266142824354,
+      "loss": 1.9435,
+      "step": 5600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004697604879514853,
+      "loss": 1.9145,
+      "step": 5610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004683085704438451,
+      "loss": 1.9975,
+      "step": 5620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00046685692119540357,
+      "loss": 1.9696,
+      "step": 5630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00046540555249397056,
+      "loss": 1.9598,
+      "step": 5640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00046395447662498117,
+      "loss": 1.9366,
+      "step": 5650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00046250370587139206,
+      "loss": 1.9748,
+      "step": 5660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00046105325251357695,
+      "loss": 1.9125,
+      "step": 5670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00045960312882922306,
+      "loss": 1.9595,
+      "step": 5680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00045815334709322665,
+      "loss": 1.9369,
+      "step": 5690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004567039195775899,
+      "loss": 1.9564,
+      "step": 5700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000455254858551316,
+      "loss": 1.9274,
+      "step": 5710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00045380617628030693,
+      "loss": 1.9366,
+      "step": 5720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00045235788502725777,
+      "loss": 1.9446,
+      "step": 5730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00045090999705155383,
+      "loss": 1.9236,
+      "step": 5740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004494625246091672,
+      "loss": 1.9397,
+      "step": 5750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00044801547995255194,
+      "loss": 1.9849,
+      "step": 5760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00044656887533054193,
+      "loss": 1.9367,
+      "step": 5770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004451227229882455,
+      "loss": 1.9451,
+      "step": 5780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004436770351669429,
+      "loss": 1.9578,
+      "step": 5790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004422318241039821,
+      "loss": 1.9095,
+      "step": 5800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00044078710203267576,
+      "loss": 1.9337,
+      "step": 5810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004393428811821971,
+      "loss": 1.9904,
+      "step": 5820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00043789917377747695,
+      "loss": 1.9569,
+      "step": 5830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004364559920390996,
+      "loss": 1.9423,
+      "step": 5840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00043501334818320004,
+      "loss": 2.0013,
+      "step": 5850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004335712544213597,
+      "loss": 1.9585,
+      "step": 5860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004321297229605044,
+      "loss": 1.9148,
+      "step": 5870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00043068876600279964,
+      "loss": 1.9558,
+      "step": 5880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004292483957455482,
+      "loss": 1.9899,
+      "step": 5890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00042780862438108655,
+      "loss": 1.9597,
+      "step": 5900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00042636946409668123,
+      "loss": 1.9603,
+      "step": 5910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00042493092707442717,
+      "loss": 1.9771,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004234930254911427,
+      "loss": 1.9416,
+      "step": 5930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00042205577151826747,
+      "loss": 1.9417,
+      "step": 5940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00042061917732175966,
+      "loss": 1.9636,
+      "step": 5950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000419183255061992,
+      "loss": 1.966,
+      "step": 5960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004177480168936504,
+      "loss": 1.9885,
+      "step": 5970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00041631347496562913,
+      "loss": 1.9525,
+      "step": 5980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004148796414209297,
+      "loss": 1.9809,
+      "step": 5990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004134465283965568,
+      "loss": 1.8982,
+      "step": 6000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00041201414802341643,
+      "loss": 1.8978,
+      "step": 6010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00041058251242621295,
+      "loss": 1.9406,
+      "step": 6020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004091516337233464,
+      "loss": 1.919,
+      "step": 6030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00040772152402680973,
+      "loss": 1.9897,
+      "step": 6040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00040629219544208644,
+      "loss": 1.9473,
+      "step": 6050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00040486366006804817,
+      "loss": 1.9589,
+      "step": 6060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00040343592999685264,
+      "loss": 1.8972,
+      "step": 6070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004020090173138404,
+      "loss": 1.9611,
+      "step": 6080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00040058293409743316,
+      "loss": 1.9164,
+      "step": 6090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039915769241903175,
+      "loss": 1.939,
+      "step": 6100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003977333043429129,
+      "loss": 1.9233,
+      "step": 6110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003963097819261288,
+      "loss": 1.9742,
+      "step": 6120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003948871372184033,
+      "loss": 1.9719,
+      "step": 6130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003934653822620309,
+      "loss": 1.9861,
+      "step": 6140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003920445290917747,
+      "loss": 1.9514,
+      "step": 6150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003906245897347643,
+      "loss": 2.0047,
+      "step": 6160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003892055762103943,
+      "loss": 1.934,
+      "step": 6170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003877875005302223,
+      "loss": 1.9619,
+      "step": 6180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00038637037469786727,
+      "loss": 1.9085,
+      "step": 6190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00038495421070890796,
+      "loss": 1.9222,
+      "step": 6200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003835390205507812,
+      "loss": 1.9409,
+      "step": 6210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00038212481620268125,
+      "loss": 1.9439,
+      "step": 6220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000380711609635457,
+      "loss": 1.974,
+      "step": 6230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000379299412811512,
+      "loss": 1.9141,
+      "step": 6240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00037788823768470227,
+      "loss": 1.8982,
+      "step": 6250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003764780962002352,
+      "loss": 1.9384,
+      "step": 6260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003750690002945695,
+      "loss": 1.9174,
+      "step": 6270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00037366096189531253,
+      "loss": 1.927,
+      "step": 6280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003722539929211205,
+      "loss": 1.9449,
+      "step": 6290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003708481052815973,
+      "loss": 2.008,
+      "step": 6300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003694433108771933,
+      "loss": 1.9345,
+      "step": 6310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00036803962159910543,
+      "loss": 1.8619,
+      "step": 6320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003666370493291757,
+      "loss": 1.9612,
+      "step": 6330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003652356059397909,
+      "loss": 1.9371,
+      "step": 6340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00036383530329378225,
+      "loss": 1.9193,
+      "step": 6350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00036243615324432494,
+      "loss": 1.9509,
+      "step": 6360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00036103816763483775,
+      "loss": 1.9517,
+      "step": 6370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00035964135829888287,
+      "loss": 1.9141,
+      "step": 6380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00035824573706006547,
+      "loss": 1.9985,
+      "step": 6390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00035685131573193386,
+      "loss": 1.9791,
+      "step": 6400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00035545810611787925,
+      "loss": 1.8865,
+      "step": 6410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003540661200110367,
+      "loss": 1.9409,
+      "step": 6420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00035267536919418395,
+      "loss": 1.9928,
+      "step": 6430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00035128586543964243,
+      "loss": 1.9149,
+      "step": 6440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00034989762050917784,
+      "loss": 1.9837,
+      "step": 6450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003485106461538999,
+      "loss": 1.9142,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00034712495411416385,
+      "loss": 1.9284,
+      "step": 6470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00034574055611946984,
+      "loss": 1.8848,
+      "step": 6480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00034435746388836457,
+      "loss": 1.9295,
+      "step": 6490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00034297568912834177,
+      "loss": 1.9635,
+      "step": 6500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003415952435357431,
+      "loss": 1.9794,
+      "step": 6510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00034021613879565944,
+      "loss": 1.9347,
+      "step": 6520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00033883838658183154,
+      "loss": 1.9282,
+      "step": 6530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00033746199855655125,
+      "loss": 1.9973,
+      "step": 6540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000336086986370563,
+      "loss": 1.9718,
+      "step": 6550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003347133616629652,
+      "loss": 1.9157,
+      "step": 6560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00033334113606111194,
+      "loss": 1.9528,
+      "step": 6570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003319703211805135,
+      "loss": 1.9369,
+      "step": 6580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00033060092862473967,
+      "loss": 1.9552,
+      "step": 6590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00032923296998532003,
+      "loss": 1.9581,
+      "step": 6600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003278664568416465,
+      "loss": 1.9406,
+      "step": 6610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003265014007608758,
+      "loss": 1.9672,
+      "step": 6620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003251378132978304,
+      "loss": 1.9166,
+      "step": 6630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00032377570599490164,
+      "loss": 1.9356,
+      "step": 6640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00032241509038195165,
+      "loss": 1.9092,
+      "step": 6650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003210559779762157,
+      "loss": 1.9343,
+      "step": 6660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00031969838028220513,
+      "loss": 1.9796,
+      "step": 6670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00031834230879160947,
+      "loss": 1.8932,
+      "step": 6680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003169877749831993,
+      "loss": 1.9279,
+      "step": 6690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000315634790322729,
+      "loss": 1.9444,
+      "step": 6700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00031428336626284037,
+      "loss": 1.9296,
+      "step": 6710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00031293351424296467,
+      "loss": 1.9341,
+      "step": 6720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00031158524568922663,
+      "loss": 1.9589,
+      "step": 6730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00031023857201434703,
+      "loss": 1.8997,
+      "step": 6740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003088935046175466,
+      "loss": 1.9417,
+      "step": 6750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003075500548844492,
+      "loss": 1.9058,
+      "step": 6760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003062082341869862,
+      "loss": 1.9511,
+      "step": 6770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00030486805388329894,
+      "loss": 1.922,
+      "step": 6780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00030352952531764363,
+      "loss": 1.9634,
+      "step": 6790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003021926598202949,
+      "loss": 1.9555,
+      "step": 6800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003008574687074498,
+      "loss": 1.9254,
+      "step": 6810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00029952396328113265,
+      "loss": 1.9211,
+      "step": 6820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029819215482909846,
+      "loss": 1.9504,
+      "step": 6830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029686205462473784,
+      "loss": 1.9174,
+      "step": 6840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029553367392698177,
+      "loss": 1.9368,
+      "step": 6850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029420702398020595,
+      "loss": 1.8939,
+      "step": 6860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029288211601413586,
+      "loss": 1.9097,
+      "step": 6870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029155896124375156,
+      "loss": 1.9885,
+      "step": 6880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002902375708691925,
+      "loss": 1.9752,
+      "step": 6890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002889179560756634,
+      "loss": 1.9321,
+      "step": 6900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002876001280333391,
+      "loss": 1.9345,
+      "step": 6910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002862840978972698,
+      "loss": 1.9192,
+      "step": 6920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002849698768072877,
+      "loss": 1.9499,
+      "step": 6930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002836574758879107,
+      "loss": 1.9217,
+      "step": 6940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002823469062482503,
+      "loss": 1.9318,
+      "step": 6950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002810381789819167,
+      "loss": 1.9422,
+      "step": 6960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002797313051669247,
+      "loss": 1.9562,
+      "step": 6970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002784262958656003,
+      "loss": 1.9189,
+      "step": 6980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00027712316212448663,
+      "loss": 1.9453,
+      "step": 6990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00027582191497425114,
+      "loss": 1.9335,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000274522565429591,
+      "loss": 1.9724,
+      "step": 7010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002732251244891421,
+      "loss": 1.9236,
+      "step": 7020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00027192960313538307,
+      "loss": 1.9407,
+      "step": 7030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002706360123345445,
+      "loss": 1.8948,
+      "step": 7040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00026934436303651554,
+      "loss": 1.9354,
+      "step": 7050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00026805466617475007,
+      "loss": 1.955,
+      "step": 7060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002667669326661767,
+      "loss": 1.9595,
+      "step": 7070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002654811734111033,
+      "loss": 1.9766,
+      "step": 7080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00026419739929312713,
+      "loss": 1.9626,
+      "step": 7090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002629156211790417,
+      "loss": 1.9096,
+      "step": 7100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00026163584991874413,
+      "loss": 1.9522,
+      "step": 7110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00026035809634514543,
+      "loss": 1.9387,
+      "step": 7120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025908237127407626,
+      "loss": 1.8788,
+      "step": 7130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002578086855041973,
+      "loss": 1.9326,
+      "step": 7140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002565370498169064,
+      "loss": 1.92,
+      "step": 7150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002552674749762487,
+      "loss": 1.9131,
+      "step": 7160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002539999717288246,
+      "loss": 1.9263,
+      "step": 7170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025286099888629365,
+      "loss": 1.919,
+      "step": 7180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000251597461210021,
+      "loss": 1.9281,
+      "step": 7190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025033602619263507,
+      "loss": 1.9831,
+      "step": 7200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024907670451183454,
+      "loss": 1.9535,
+      "step": 7210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024781950682742983,
+      "loss": 1.8798,
+      "step": 7220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024656444378125204,
+      "loss": 1.9124,
+      "step": 7230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024531152599706346,
+      "loss": 1.9472,
+      "step": 7240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002440607640804668,
+      "loss": 1.9697,
+      "step": 7250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024281216861881633,
+      "loss": 1.938,
+      "step": 7260
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024156575018112787,
+      "loss": 1.9035,
+      "step": 7270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024032151931798918,
+      "loss": 1.9379,
+      "step": 7280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002390794865614711,
+      "loss": 1.9176,
+      "step": 7290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002378396624250375,
+      "loss": 1.9392,
+      "step": 7300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002366020574034576,
+      "loss": 1.9235,
+      "step": 7310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023536668197271588,
+      "loss": 1.939,
+      "step": 7320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002341335465899243,
+      "loss": 1.889,
+      "step": 7330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023290266169323354,
+      "loss": 1.9247,
+      "step": 7340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002316740377017438,
+      "loss": 1.935,
+      "step": 7350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002304476850154183,
+      "loss": 1.9198,
+      "step": 7360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00022922361401499325,
+      "loss": 1.9068,
+      "step": 7370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00022800183506189238,
+      "loss": 1.9505,
+      "step": 7380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00022678235849813645,
+      "loss": 1.9459,
+      "step": 7390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00022556519464625807,
+      "loss": 1.9302,
+      "step": 7400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00022435035380921321,
+      "loss": 1.9345,
+      "step": 7410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002231378462702935,
+      "loss": 1.9304,
+      "step": 7420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022192768229304107,
+      "loss": 1.9457,
+      "step": 7430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002207198721211593,
+      "loss": 1.9395,
+      "step": 7440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021951442597842785,
+      "loss": 1.9867,
+      "step": 7450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021831135406861558,
+      "loss": 1.9217,
+      "step": 7460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00021711066657539342,
+      "loss": 1.991,
+      "step": 7470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021591237366225008,
+      "loss": 1.9492,
+      "step": 7480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021471648547240365,
+      "loss": 1.9376,
+      "step": 7490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021352301212871762,
+      "loss": 1.9436,
+      "step": 7500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021233196373361397,
+      "loss": 1.9267,
+      "step": 7510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021114335036898852,
+      "loss": 1.9483,
+      "step": 7520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002099571820961252,
+      "loss": 1.9485,
+      "step": 7530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020877346895561082,
+      "loss": 1.9607,
+      "step": 7540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020759222096725034,
+      "loss": 1.9665,
+      "step": 7550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002064134481299814,
+      "loss": 1.9202,
+      "step": 7560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020523716042179075,
+      "loss": 1.9375,
+      "step": 7570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00020406336779962888,
+      "loss": 1.9406,
+      "step": 7580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020289208019932635,
+      "loss": 1.9357,
+      "step": 7590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002017233075355089,
+      "loss": 1.9432,
+      "step": 7600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002005570597015145,
+      "loss": 1.962,
+      "step": 7610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019939334656930892,
+      "loss": 1.9237,
+      "step": 7620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00019823217798940247,
+      "loss": 1.9688,
+      "step": 7630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019707356379076668,
+      "loss": 1.9524,
+      "step": 7640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019591751378075035,
+      "loss": 1.981,
+      "step": 7650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019476403774499774,
+      "loss": 1.9405,
+      "step": 7660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001936131454473649,
+      "loss": 1.9,
+      "step": 7670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019246484662983733,
+      "loss": 1.8852,
+      "step": 7680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00019131915101244752,
+      "loss": 1.9447,
+      "step": 7690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019017606829319205,
+      "loss": 1.9221,
+      "step": 7700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00018903560814795086,
+      "loss": 1.9481,
+      "step": 7710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00018789778023040355,
+      "loss": 1.8919,
+      "step": 7720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001867625941719499,
+      "loss": 1.94,
+      "step": 7730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00018563005958162587,
+      "loss": 1.9089,
+      "step": 7740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00018450018604602414,
+      "loss": 1.9105,
+      "step": 7750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001833729831292123,
+      "loss": 1.9568,
+      "step": 7760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00018224846037265125,
+      "loss": 1.9156,
+      "step": 7770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001811266272951162,
+      "loss": 1.9445,
+      "step": 7780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001800074933926138,
+      "loss": 1.8871,
+      "step": 7790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017889106813830363,
+      "loss": 1.9684,
+      "step": 7800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001777773609824173,
+      "loss": 1.9411,
+      "step": 7810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017666638135217783,
+      "loss": 1.9295,
+      "step": 7820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001755581386517216,
+      "loss": 1.9499,
+      "step": 7830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001744526422620165,
+      "loss": 1.9499,
+      "step": 7840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017334990154078446,
+      "loss": 1.9262,
+      "step": 7850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017224992582242076,
+      "loss": 1.9554,
+      "step": 7860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017115272441791635,
+      "loss": 1.9685,
+      "step": 7870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000170058306614778,
+      "loss": 1.9021,
+      "step": 7880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016896668167695018,
+      "loss": 1.8804,
+      "step": 7890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016787785884473665,
+      "loss": 1.9164,
+      "step": 7900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016679184733472163,
+      "loss": 1.9409,
+      "step": 7910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016570865633969283,
+      "loss": 1.9329,
+      "step": 7920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016462829502856286,
+      "loss": 1.9518,
+      "step": 7930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016355077254629196,
+      "loss": 1.9494,
+      "step": 7940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016247609801381003,
+      "loss": 1.9376,
+      "step": 7950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016140428052794042,
+      "loss": 1.9312,
+      "step": 7960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016033532916132203,
+      "loss": 1.9528,
+      "step": 7970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015926925296233296,
+      "loss": 1.9698,
+      "step": 7980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015820606095501405,
+      "loss": 1.936,
+      "step": 7990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015714576213899167,
+      "loss": 1.962,
+      "step": 8000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015608836548940252,
+      "loss": 1.9439,
+      "step": 8010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015503387995681732,
+      "loss": 1.9274,
+      "step": 8020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001539823144671648,
+      "loss": 1.9029,
+      "step": 8030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015293367792165664,
+      "loss": 1.928,
+      "step": 8040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015188797919671115,
+      "loss": 1.9405,
+      "step": 8050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015084522714387967,
+      "loss": 1.909,
+      "step": 8060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014980543058976977,
+      "loss": 1.9286,
+      "step": 8070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014876859833597278,
+      "loss": 1.9358,
+      "step": 8080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014773473915898695,
+      "loss": 1.9073,
+      "step": 8090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014670386181014477,
+      "loss": 1.9044,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014567597501553847,
+      "loss": 1.9133,
+      "step": 8110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001446510874759454,
+      "loss": 1.9583,
+      "step": 8120
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014362920786675616,
+      "loss": 1.9381,
+      "step": 8130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014261034483789892,
+      "loss": 1.951,
+      "step": 8140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014159450701376803,
+      "loss": 1.9806,
+      "step": 8150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014058170299315027,
+      "loss": 1.9101,
+      "step": 8160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013957194134915147,
+      "loss": 1.9122,
+      "step": 8170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013856523062912569,
+      "loss": 1.91,
+      "step": 8180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001375615793546006,
+      "loss": 1.9182,
+      "step": 8190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013656099602120752,
+      "loss": 1.9312,
+      "step": 8200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013556348909860762,
+      "loss": 1.9327,
+      "step": 8210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013456906703042165,
+      "loss": 1.9184,
+      "step": 8220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013357773823415782,
+      "loss": 1.8876,
+      "step": 8230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013258951110114054,
+      "loss": 1.927,
+      "step": 8240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001316043939964398,
+      "loss": 1.9525,
+      "step": 8250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013062239525879943,
+      "loss": 1.8933,
+      "step": 8260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001296435232005677,
+      "loss": 1.9433,
+      "step": 8270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001286677861076263,
+      "loss": 1.9306,
+      "step": 8280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00012769519223932034,
+      "loss": 1.8962,
+      "step": 8290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00012672574982838802,
+      "loss": 1.9322,
+      "step": 8300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00012575946708089176,
+      "loss": 1.9191,
+      "step": 8310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001247963521761481,
+      "loss": 1.9308,
+      "step": 8320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00012383641326665874,
+      "loss": 1.8876,
+      "step": 8330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00012287965847804162,
+      "loss": 1.9103,
+      "step": 8340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012192609590896142,
+      "loss": 1.8804,
+      "step": 8350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012097573363106218,
+      "loss": 1.944,
+      "step": 8360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012002857968889797,
+      "loss": 1.9259,
+      "step": 8370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00011908464209986541,
+      "loss": 1.9421,
+      "step": 8380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00011814392885413567,
+      "loss": 1.9251,
+      "step": 8390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011720644791458623,
+      "loss": 1.9106,
+      "step": 8400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011627220721673459,
+      "loss": 1.8928,
+      "step": 8410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011534121466866986,
+      "loss": 1.9323,
+      "step": 8420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011441347815098735,
+      "loss": 1.9238,
+      "step": 8430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001134890055167201,
+      "loss": 1.9831,
+      "step": 8440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011256780459127374,
+      "loss": 1.85,
+      "step": 8450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001116498831723598,
+      "loss": 1.9391,
+      "step": 8460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011073524902992921,
+      "loss": 1.9071,
+      "step": 8470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010982390990610791,
+      "loss": 1.9538,
+      "step": 8480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010891587351512949,
+      "loss": 1.8972,
+      "step": 8490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010801114754327135,
+      "loss": 1.9184,
+      "step": 8500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010710973964878906,
+      "loss": 1.9567,
+      "step": 8510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010621165746185113,
+      "loss": 1.9755,
+      "step": 8520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001053169085844758,
+      "loss": 1.9256,
+      "step": 8530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001044255005904648,
+      "loss": 1.9742,
+      "step": 8540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001035374410253408,
+      "loss": 1.9562,
+      "step": 8550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001026527374062825,
+      "loss": 1.9432,
+      "step": 8560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010177139722206191,
+      "loss": 1.9007,
+      "step": 8570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010089342793298028,
+      "loss": 1.9834,
+      "step": 8580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010001883697080504,
+      "loss": 1.9063,
+      "step": 8590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.914763173870738e-05,
+      "loss": 1.9063,
+      "step": 8600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.827981961119869e-05,
+      "loss": 1.9149,
+      "step": 8610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.741540793406911e-05,
+      "loss": 1.9096,
+      "step": 8620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.655440402432464e-05,
+      "loss": 1.8657,
+      "step": 8630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.569681517012585e-05,
+      "loss": 1.9237,
+      "step": 8640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.484264863072523e-05,
+      "loss": 1.9739,
+      "step": 8650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.399191163640675e-05,
+      "loss": 1.9199,
+      "step": 8660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314461138842406e-05,
+      "loss": 1.9639,
+      "step": 8670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.230075505893987e-05,
+      "loss": 1.9391,
+      "step": 8680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.146034979096496e-05,
+      "loss": 1.9326,
+      "step": 8690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.06234026982976e-05,
+      "loss": 1.9501,
+      "step": 8700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.9789920865464e-05,
+      "loss": 1.9041,
+      "step": 8710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.895991134765758e-05,
+      "loss": 1.931,
+      "step": 8720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.813338117067965e-05,
+      "loss": 1.9589,
+      "step": 8730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.731033733088012e-05,
+      "loss": 1.9226,
+      "step": 8740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.649078679509742e-05,
+      "loss": 1.911,
+      "step": 8750
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.567473650060082e-05,
+      "loss": 1.9641,
+      "step": 8760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.486219335503026e-05,
+      "loss": 1.8924,
+      "step": 8770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.405316423633958e-05,
+      "loss": 1.9583,
+      "step": 8780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.324765599273649e-05,
+      "loss": 1.9399,
+      "step": 8790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.244567544262588e-05,
+      "loss": 1.9362,
+      "step": 8800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.164722937455177e-05,
+      "loss": 1.9283,
+      "step": 8810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.08523245471392e-05,
+      "loss": 1.9588,
+      "step": 8820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.006096768903848e-05,
+      "loss": 1.9126,
+      "step": 8830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.927316549886659e-05,
+      "loss": 1.9331,
+      "step": 8840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.84889246451515e-05,
+      "loss": 1.8967,
+      "step": 8850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.77082517662756e-05,
+      "loss": 1.927,
+      "step": 8860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.693115347041884e-05,
+      "loss": 1.8978,
+      "step": 8870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.615763633550421e-05,
+      "loss": 1.9142,
+      "step": 8880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.538770690914015e-05,
+      "loss": 1.9001,
+      "step": 8890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.462137170856698e-05,
+      "loss": 1.8979,
+      "step": 8900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.385863722060022e-05,
+      "loss": 1.892,
+      "step": 8910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.309950990157677e-05,
+      "loss": 1.9496,
+      "step": 8920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.234399617729976e-05,
+      "loss": 1.8879,
+      "step": 8930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.159210244298414e-05,
+      "loss": 1.921,
+      "step": 8940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.084383506320291e-05,
+      "loss": 1.9408,
+      "step": 8950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.009920037183248e-05,
+      "loss": 1.9258,
+      "step": 8960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.935820467200005e-05,
+      "loss": 1.9182,
+      "step": 8970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.862085423602955e-05,
+      "loss": 1.9668,
+      "step": 8980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.78871553053888e-05,
+      "loss": 1.898,
+      "step": 8990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.715711409063652e-05,
+      "loss": 1.9261,
+      "step": 9000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.643073677136996e-05,
+      "loss": 1.9167,
+      "step": 9010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.570802949617261e-05,
+      "loss": 1.9066,
+      "step": 9020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.49889983825619e-05,
+      "loss": 1.9091,
+      "step": 9030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.427364951693776e-05,
+      "loss": 1.8853,
+      "step": 9040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.356198895453053e-05,
+      "loss": 1.9459,
+      "step": 9050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.285402271935048e-05,
+      "loss": 1.9073,
+      "step": 9060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.214975680413621e-05,
+      "loss": 1.902,
+      "step": 9070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.144919717030417e-05,
+      "loss": 1.8812,
+      "step": 9080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.075234974789839e-05,
+      "loss": 1.8672,
+      "step": 9090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.005922043553952e-05,
+      "loss": 1.9297,
+      "step": 9100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.9369815100376037e-05,
+      "loss": 1.8775,
+      "step": 9110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.868413957803331e-05,
+      "loss": 1.8831,
+      "step": 9120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.800219967256576e-05,
+      "loss": 1.9026,
+      "step": 9130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7324001156406014e-05,
+      "loss": 1.8944,
+      "step": 9140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.664954977031733e-05,
+      "loss": 1.9154,
+      "step": 9150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.5978851223344455e-05,
+      "loss": 1.8994,
+      "step": 9160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.531191119276513e-05,
+      "loss": 1.9397,
+      "step": 9170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.4648735324042855e-05,
+      "loss": 1.9416,
+      "step": 9180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.398932923077793e-05,
+      "loss": 1.9207,
+      "step": 9190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.333369849466091e-05,
+      "loss": 1.9061,
+      "step": 9200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.268184866542502e-05,
+      "loss": 1.8861,
+      "step": 9210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.2033785260798716e-05,
+      "loss": 1.9433,
+      "step": 9220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.138951376646012e-05,
+      "loss": 1.9158,
+      "step": 9230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.074903963598926e-05,
+      "loss": 1.8898,
+      "step": 9240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.0112368290822885e-05,
+      "loss": 1.9041,
+      "step": 9250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9479505120207934e-05,
+      "loss": 1.9605,
+      "step": 9260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.885045548115646e-05,
+      "loss": 1.8975,
+      "step": 9270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.822522469839996e-05,
+      "loss": 1.9396,
+      "step": 9280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.760381806434427e-05,
+      "loss": 1.9263,
+      "step": 9290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.698624083902503e-05,
+      "loss": 1.948,
+      "step": 9300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.637249825006273e-05,
+      "loss": 1.9206,
+      "step": 9310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5762595492618906e-05,
+      "loss": 1.9129,
+      "step": 9320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.515653772935207e-05,
+      "loss": 1.9036,
+      "step": 9330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.455433009037385e-05,
+      "loss": 1.928,
+      "step": 9340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.395597767320536e-05,
+      "loss": 1.9427,
+      "step": 9350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.336148554273467e-05,
+      "loss": 1.8882,
+      "step": 9360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.277085873117354e-05,
+      "loss": 1.9412,
+      "step": 9370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.218410223801472e-05,
+      "loss": 1.9014,
+      "step": 9380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1601221029989976e-05,
+      "loss": 1.9056,
+      "step": 9390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.102222004102768e-05,
+      "loss": 1.9665,
+      "step": 9400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.04471041722112e-05,
+      "loss": 1.8911,
+      "step": 9410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.987587829173772e-05,
+      "loss": 1.885,
+      "step": 9420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.930854723487642e-05,
+      "loss": 1.8958,
+      "step": 9430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.874511580392814e-05,
+      "loss": 1.9197,
+      "step": 9440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.818558876818418e-05,
+      "loss": 1.9174,
+      "step": 9450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.762997086388653e-05,
+      "loss": 1.9321,
+      "step": 9460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.707826679418702e-05,
+      "loss": 1.8835,
+      "step": 9470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.653048122910862e-05,
+      "loss": 1.9083,
+      "step": 9480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.598661880550452e-05,
+      "loss": 1.9082,
+      "step": 9490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.544668412702007e-05,
+      "loss": 1.9696,
+      "step": 9500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.491068176405326e-05,
+      "loss": 1.9232,
+      "step": 9510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.43786162537158e-05,
+      "loss": 1.8944,
+      "step": 9520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.385049209979568e-05,
+      "loss": 1.9341,
+      "step": 9530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3326313772717667e-05,
+      "loss": 1.9,
+      "step": 9540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2806085709506716e-05,
+      "loss": 1.9157,
+      "step": 9550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.228981231374972e-05,
+      "loss": 1.9287,
+      "step": 9560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.177749795555812e-05,
+      "loss": 1.9375,
+      "step": 9570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.126914697153182e-05,
+      "loss": 1.9171,
+      "step": 9580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.076476366472131e-05,
+      "loss": 1.8778,
+      "step": 9590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0264352304592123e-05,
+      "loss": 1.919,
+      "step": 9600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9767917126987952e-05,
+      "loss": 1.891,
+      "step": 9610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.927546233409567e-05,
+      "loss": 1.921,
+      "step": 9620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8786992094409102e-05,
+      "loss": 1.9178,
+      "step": 9630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8302510542693782e-05,
+      "loss": 1.8917,
+      "step": 9640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7822021779952467e-05,
+      "loss": 1.9141,
+      "step": 9650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7345529873389618e-05,
+      "loss": 1.8737,
+      "step": 9660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.687303885637765e-05,
+      "loss": 1.9349,
+      "step": 9670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.6404552728422625e-05,
+      "loss": 1.919,
+      "step": 9680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.5940075455130217e-05,
+      "loss": 1.9411,
+      "step": 9690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.547961096817214e-05,
+      "loss": 1.9616,
+      "step": 9700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5023163165253272e-05,
+      "loss": 1.9431,
+      "step": 9710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4570735910078078e-05,
+      "loss": 1.9488,
+      "step": 9720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4122333032318367e-05,
+      "loss": 1.9376,
+      "step": 9730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.3677958327580683e-05,
+      "loss": 1.9038,
+      "step": 9740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3237615557374137e-05,
+      "loss": 1.91,
+      "step": 9750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2801308449078684e-05,
+      "loss": 1.9058,
+      "step": 9760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2369040695913445e-05,
+      "loss": 1.9275,
+      "step": 9770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1940815956905612e-05,
+      "loss": 1.9646,
+      "step": 9780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1516637856859422e-05,
+      "loss": 1.9127,
+      "step": 9790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.109650998632512e-05,
+      "loss": 1.9564,
+      "step": 9800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.068043590156926e-05,
+      "loss": 1.929,
+      "step": 9810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0268419124544078e-05,
+      "loss": 1.9185,
+      "step": 9820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.986046314285772e-05,
+      "loss": 1.8714,
+      "step": 9830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9456571409744938e-05,
+      "loss": 1.9149,
+      "step": 9840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9056747344037616e-05,
+      "loss": 1.8945,
+      "step": 9850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8660994330136184e-05,
+      "loss": 1.9634,
+      "step": 9860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8269315717980518e-05,
+      "loss": 1.9449,
+      "step": 9870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7881714823021976e-05,
+      "loss": 1.8911,
+      "step": 9880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7498194926194978e-05,
+      "loss": 1.9169,
+      "step": 9890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7118759273889615e-05,
+      "loss": 1.8699,
+      "step": 9900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.674341107792382e-05,
+      "loss": 1.9187,
+      "step": 9910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6372153515516463e-05,
+      "loss": 1.8754,
+      "step": 9920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.600498972926029e-05,
+      "loss": 1.9913,
+      "step": 9930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5641922827095256e-05,
+      "loss": 1.9162,
+      "step": 9940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.528295588228251e-05,
+      "loss": 1.9475,
+      "step": 9950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4928091933378018e-05,
+      "loss": 1.8907,
+      "step": 9960
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4577333984207186e-05,
+      "loss": 1.9031,
+      "step": 9970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4230685003839117e-05,
+      "loss": 1.8969,
+      "step": 9980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3888147926561733e-05,
+      "loss": 1.9272,
+      "step": 9990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3549725651856804e-05,
+      "loss": 1.9313,
+      "step": 10000
+    }
+  ],
+  "max_steps": 10798,
+  "num_train_epochs": 2,
+  "total_flos": 9.488737619882803e+18,
+  "trial_name": null,
+  "trial_params": null
+}