diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14211 @@
+{
+  "best_metric": 2.882770538330078,
+  "best_model_checkpoint": "./binders_output/checkpoint-23645",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 23645,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5936680271799274e-05,
+      "loss": 4.7069,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.187336054359855e-05,
+      "loss": 2.9752,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.781004081539782e-05,
+      "loss": 2.9183,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.37467210871971e-05,
+      "loss": 2.9189,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.968340135899636e-05,
+      "loss": 2.9121,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.562008163079564e-05,
+      "loss": 2.8856,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011155676190259492,
+      "loss": 2.9224,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001274934421743942,
+      "loss": 2.9169,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014343012244619346,
+      "loss": 2.9056,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015936680271799272,
+      "loss": 2.9091,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000175303482989792,
+      "loss": 2.9047,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019124016326159128,
+      "loss": 2.9065,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020717684353339054,
+      "loss": 2.8929,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022311352380518983,
+      "loss": 2.8963,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023905020407698907,
+      "loss": 2.8896,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002549868843487884,
+      "loss": 2.9051,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002709235646205876,
+      "loss": 2.885,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002868602448923869,
+      "loss": 2.8806,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003027969251641862,
+      "loss": 2.8729,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031873360543598544,
+      "loss": 2.8872,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003346702857077847,
+      "loss": 2.8932,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000350606965979584,
+      "loss": 2.8929,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036654364625138324,
+      "loss": 2.8935,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038248032652318255,
+      "loss": 2.8887,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003984170067949818,
+      "loss": 2.8917,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004143536870667811,
+      "loss": 2.8882,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004302903673385804,
+      "loss": 2.8966,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044622704761037966,
+      "loss": 2.8872,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004621637278821789,
+      "loss": 2.8864,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047810040815397814,
+      "loss": 2.8888,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004940370884257775,
+      "loss": 2.895,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005099737686975768,
+      "loss": 2.8993,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000525910448969376,
+      "loss": 2.8755,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005418471292411752,
+      "loss": 2.8708,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005577838095129746,
+      "loss": 2.8905,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005737204897847738,
+      "loss": 2.8792,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000589657170056573,
+      "loss": 2.9267,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006055938503283724,
+      "loss": 2.8936,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006215305306001717,
+      "loss": 2.8936,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006374672108719709,
+      "loss": 2.8878,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006534038911437702,
+      "loss": 2.8783,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006693405714155694,
+      "loss": 2.9091,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006852772516873687,
+      "loss": 2.8935,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000701213931959168,
+      "loss": 2.8923,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007171506122309673,
+      "loss": 2.9048,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007330872925027665,
+      "loss": 2.889,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007490239727745659,
+      "loss": 2.8912,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007649606530463651,
+      "loss": 2.8993,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007808973333181643,
+      "loss": 2.8933,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007968340135899636,
+      "loss": 2.9004,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007981171972957405,
+      "loss": 2.8928,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007977722147164036,
+      "loss": 2.9058,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007974272321370668,
+      "loss": 2.8973,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00079708224955773,
+      "loss": 2.8874,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007967372669783931,
+      "loss": 2.8941,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007963922843990563,
+      "loss": 2.8782,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007960473018197195,
+      "loss": 2.8937,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007957023192403827,
+      "loss": 2.8891,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007953573366610458,
+      "loss": 2.9329,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000795012354081709,
+      "loss": 2.8897,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007946673715023723,
+      "loss": 2.8943,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007943223889230354,
+      "loss": 2.8994,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007939774063436986,
+      "loss": 2.916,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007936324237643618,
+      "loss": 2.9122,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007932874411850249,
+      "loss": 2.9079,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007929424586056881,
+      "loss": 2.9086,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007925974760263513,
+      "loss": 2.8991,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007922524934470144,
+      "loss": 2.9047,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007919075108676776,
+      "loss": 2.908,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007915625282883408,
+      "loss": 2.9136,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000791217545709004,
+      "loss": 2.9071,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007908725631296671,
+      "loss": 2.9109,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007905275805503303,
+      "loss": 2.9071,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007901825979709935,
+      "loss": 2.8898,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007898376153916566,
+      "loss": 2.966,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007894926328123198,
+      "loss": 2.8961,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000789147650232983,
+      "loss": 2.9039,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007888026676536461,
+      "loss": 2.8995,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007884576850743094,
+      "loss": 2.8949,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007881127024949726,
+      "loss": 2.9042,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007877677199156357,
+      "loss": 2.9026,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0007874227373362989,
+      "loss": 2.9071,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007870777547569621,
+      "loss": 2.8943,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007867327721776253,
+      "loss": 2.8971,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007863877895982884,
+      "loss": 2.9048,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007860428070189516,
+      "loss": 2.947,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007856978244396148,
+      "loss": 2.8978,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007853528418602779,
+      "loss": 2.9067,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007850078592809411,
+      "loss": 2.904,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007846628767016043,
+      "loss": 2.9054,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007843178941222675,
+      "loss": 2.9128,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007839729115429306,
+      "loss": 2.9172,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007836279289635938,
+      "loss": 2.9007,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000783282946384257,
+      "loss": 2.9103,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007829379638049202,
+      "loss": 2.9102,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007825929812255835,
+      "loss": 2.9068,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007822479986462466,
+      "loss": 2.9034,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007819030160669098,
+      "loss": 2.9069,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000781558033487573,
+      "loss": 2.9096,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007812130509082362,
+      "loss": 2.9056,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007808680683288993,
+      "loss": 2.8998,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007805230857495625,
+      "loss": 2.9027,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007801781031702257,
+      "loss": 2.8941,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007798331205908888,
+      "loss": 2.921,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000779488138011552,
+      "loss": 2.9134,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007791431554322152,
+      "loss": 2.9011,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007787981728528783,
+      "loss": 2.9037,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007784531902735415,
+      "loss": 2.9024,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007781082076942047,
+      "loss": 2.8979,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007777632251148679,
+      "loss": 2.9087,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000777418242535531,
+      "loss": 2.9102,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007770732599561942,
+      "loss": 2.9073,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007767282773768575,
+      "loss": 2.8987,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007763832947975206,
+      "loss": 2.9077,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007760383122181838,
+      "loss": 2.8981,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000775693329638847,
+      "loss": 2.9141,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007753483470595101,
+      "loss": 2.8993,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007750033644801733,
+      "loss": 2.9133,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007746583819008365,
+      "loss": 2.9077,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007743133993214996,
+      "loss": 2.8985,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007739684167421628,
+      "loss": 2.9066,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000773623434162826,
+      "loss": 2.8988,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007732784515834892,
+      "loss": 2.9092,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007729334690041523,
+      "loss": 2.9066,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007725884864248155,
+      "loss": 2.8942,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007722435038454787,
+      "loss": 2.9075,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007718985212661418,
+      "loss": 2.9086,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000771553538686805,
+      "loss": 2.9021,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007712085561074682,
+      "loss": 2.8972,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007708635735281313,
+      "loss": 2.899,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007705185909487945,
+      "loss": 2.904,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007701736083694578,
+      "loss": 2.9083,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007698286257901209,
+      "loss": 2.8974,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007694836432107841,
+      "loss": 2.9026,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007691386606314473,
+      "loss": 2.9046,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007687936780521105,
+      "loss": 2.9027,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007684486954727736,
+      "loss": 2.9165,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007681037128934368,
+      "loss": 2.9178,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007677587303141,
+      "loss": 2.9157,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007674137477347631,
+      "loss": 2.8971,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007670687651554263,
+      "loss": 2.9068,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007667237825760895,
+      "loss": 2.8961,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007663787999967526,
+      "loss": 2.917,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007660338174174158,
+      "loss": 2.9164,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000765688834838079,
+      "loss": 2.9095,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007653438522587422,
+      "loss": 2.9077,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007649988696794054,
+      "loss": 2.9107,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007646538871000685,
+      "loss": 2.901,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007643089045207318,
+      "loss": 2.9129,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000763963921941395,
+      "loss": 2.908,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007636189393620582,
+      "loss": 2.9081,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007632739567827214,
+      "loss": 2.9115,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007629289742033845,
+      "loss": 2.9024,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007625839916240477,
+      "loss": 2.9145,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007622390090447109,
+      "loss": 2.9116,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000761894026465374,
+      "loss": 2.9083,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007615490438860372,
+      "loss": 2.9007,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007612040613067004,
+      "loss": 2.9011,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007608590787273635,
+      "loss": 2.8919,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007605140961480267,
+      "loss": 2.9018,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007601691135686899,
+      "loss": 2.9054,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007598241309893531,
+      "loss": 2.911,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007594791484100162,
+      "loss": 2.8993,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007591341658306794,
+      "loss": 2.9164,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007587891832513427,
+      "loss": 2.9167,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007584442006720058,
+      "loss": 2.9032,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000758099218092669,
+      "loss": 2.9178,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007577542355133322,
+      "loss": 2.9072,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007574092529339953,
+      "loss": 2.9003,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007570642703546585,
+      "loss": 2.9074,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007567192877753217,
+      "loss": 2.9146,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007563743051959848,
+      "loss": 2.9075,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000756029322616648,
+      "loss": 2.9143,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007556843400373112,
+      "loss": 2.9161,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007553393574579744,
+      "loss": 2.9089,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007549943748786375,
+      "loss": 2.8998,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007546493922993007,
+      "loss": 2.9082,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007543044097199639,
+      "loss": 2.9111,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000753959427140627,
+      "loss": 2.9164,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007536144445612902,
+      "loss": 2.9026,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007532694619819534,
+      "loss": 2.9049,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007529244794026165,
+      "loss": 2.9122,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007525794968232797,
+      "loss": 2.8985,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000752234514243943,
+      "loss": 2.9031,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007518895316646061,
+      "loss": 2.9042,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007515445490852693,
+      "loss": 2.9106,
+      "step": 1860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007511995665059325,
+      "loss": 2.9207,
+      "step": 1870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007508545839265957,
+      "loss": 2.9015,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007505096013472588,
+      "loss": 2.9159,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000750164618767922,
+      "loss": 2.9139,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007498196361885852,
+      "loss": 2.9012,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007494746536092483,
+      "loss": 2.9102,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007491296710299115,
+      "loss": 2.896,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007487846884505747,
+      "loss": 2.909,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007484397058712378,
+      "loss": 2.902,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000748094723291901,
+      "loss": 2.9045,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007477497407125642,
+      "loss": 2.8972,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007474047581332274,
+      "loss": 2.9215,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007470597755538905,
+      "loss": 2.9105,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007467147929745537,
+      "loss": 2.9027,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000746369810395217,
+      "loss": 2.9123,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007460248278158802,
+      "loss": 2.9051,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007456798452365434,
+      "loss": 2.9044,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007453348626572065,
+      "loss": 2.9032,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007449898800778697,
+      "loss": 2.9019,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007446448974985329,
+      "loss": 2.911,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007442999149191961,
+      "loss": 2.907,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007439549323398592,
+      "loss": 2.9053,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007436099497605224,
+      "loss": 2.9113,
+      "step": 2090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007432649671811856,
+      "loss": 2.9134,
+      "step": 2100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007429199846018488,
+      "loss": 2.8867,
+      "step": 2110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007425750020225119,
+      "loss": 2.9099,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007422300194431751,
+      "loss": 2.8953,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007418850368638383,
+      "loss": 2.9164,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007415400542845014,
+      "loss": 2.9145,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007411950717051646,
+      "loss": 2.9123,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007408500891258278,
+      "loss": 2.9061,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000740505106546491,
+      "loss": 2.8877,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007401601239671542,
+      "loss": 2.9,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007398151413878174,
+      "loss": 2.9122,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007394701588084805,
+      "loss": 2.8987,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007391251762291437,
+      "loss": 2.8998,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007387801936498069,
+      "loss": 2.9058,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007384352110704701,
+      "loss": 2.9145,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007380902284911332,
+      "loss": 2.9144,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007377452459117964,
+      "loss": 2.904,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007374002633324596,
+      "loss": 2.906,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007370552807531227,
+      "loss": 2.9077,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007367102981737859,
+      "loss": 2.8978,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007363653155944491,
+      "loss": 2.9142,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007360203330151122,
+      "loss": 2.9127,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007356753504357754,
+      "loss": 2.9056,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007353303678564386,
+      "loss": 2.9065,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007349853852771017,
+      "loss": 2.9075,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007346404026977649,
+      "loss": 2.9038,
+      "step": 2350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007342954201184282,
+      "loss": 2.9001,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007339504375390914,
+      "loss": 2.9085,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007336054549597545,
+      "loss": 2.9131,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007332604723804177,
+      "loss": 2.9041,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007329154898010809,
+      "loss": 2.912,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000732570507221744,
+      "loss": 2.9028,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007322255246424072,
+      "loss": 2.906,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007318805420630704,
+      "loss": 2.8932,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007315355594837335,
+      "loss": 2.8984,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007311905769043967,
+      "loss": 2.8975,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007308455943250599,
+      "loss": 2.9068,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000730500611745723,
+      "loss": 2.8922,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007301556291663862,
+      "loss": 2.9151,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007298106465870494,
+      "loss": 2.9085,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007294656640077126,
+      "loss": 2.9102,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007291206814283757,
+      "loss": 2.893,
+      "step": 2510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007287756988490389,
+      "loss": 2.916,
+      "step": 2520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007284307162697022,
+      "loss": 2.9004,
+      "step": 2530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007280857336903654,
+      "loss": 2.9056,
+      "step": 2540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007277407511110285,
+      "loss": 2.8997,
+      "step": 2550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007273957685316917,
+      "loss": 2.9095,
+      "step": 2560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007270507859523549,
+      "loss": 2.9179,
+      "step": 2570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007267058033730181,
+      "loss": 2.9103,
+      "step": 2580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007263608207936813,
+      "loss": 2.9072,
+      "step": 2590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007260158382143444,
+      "loss": 2.9025,
+      "step": 2600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007256708556350076,
+      "loss": 2.9052,
+      "step": 2610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007253258730556708,
+      "loss": 2.9051,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000724980890476334,
+      "loss": 2.9094,
+      "step": 2630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007246359078969971,
+      "loss": 2.8945,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007242909253176603,
+      "loss": 2.9023,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007239459427383235,
+      "loss": 2.9097,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007236009601589866,
+      "loss": 2.9203,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007232559775796498,
+      "loss": 2.9019,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000722910995000313,
+      "loss": 2.9033,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007225660124209761,
+      "loss": 2.908,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007222210298416394,
+      "loss": 2.9078,
+      "step": 2710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007218760472623026,
+      "loss": 2.9074,
+      "step": 2720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007215310646829657,
+      "loss": 2.902,
+      "step": 2730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007211860821036289,
+      "loss": 2.91,
+      "step": 2740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007208410995242921,
+      "loss": 2.9052,
+      "step": 2750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007204961169449553,
+      "loss": 2.8997,
+      "step": 2760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007201511343656184,
+      "loss": 2.9101,
+      "step": 2770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007198061517862816,
+      "loss": 2.8938,
+      "step": 2780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007194611692069448,
+      "loss": 2.9142,
+      "step": 2790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007191161866276079,
+      "loss": 2.9222,
+      "step": 2800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007187712040482711,
+      "loss": 2.9007,
+      "step": 2810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007184262214689343,
+      "loss": 2.8993,
+      "step": 2820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007180812388895974,
+      "loss": 2.8981,
+      "step": 2830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007177362563102606,
+      "loss": 2.9096,
+      "step": 2840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007173912737309238,
+      "loss": 2.9104,
+      "step": 2850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007170462911515869,
+      "loss": 2.8914,
+      "step": 2860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007167013085722501,
+      "loss": 2.9086,
+      "step": 2870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007163563259929134,
+      "loss": 2.8948,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007160113434135766,
+      "loss": 2.9029,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007156663608342397,
+      "loss": 2.9011,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007153213782549029,
+      "loss": 2.9083,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007149763956755661,
+      "loss": 2.9049,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007146314130962292,
+      "loss": 2.8892,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007142864305168924,
+      "loss": 2.9068,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007139414479375556,
+      "loss": 2.9161,
+      "step": 2950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007135964653582187,
+      "loss": 2.897,
+      "step": 2960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007132514827788819,
+      "loss": 2.9236,
+      "step": 2970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007129065001995451,
+      "loss": 2.9071,
+      "step": 2980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007125615176202082,
+      "loss": 2.9076,
+      "step": 2990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007122165350408714,
+      "loss": 2.9089,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007118715524615346,
+      "loss": 2.8822,
+      "step": 3010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007115265698821978,
+      "loss": 2.9114,
+      "step": 3020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007111815873028609,
+      "loss": 2.906,
+      "step": 3030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007108366047235241,
+      "loss": 2.901,
+      "step": 3040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007104916221441873,
+      "loss": 2.913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007101466395648504,
+      "loss": 2.9048,
+      "step": 3060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007098016569855137,
+      "loss": 2.9206,
+      "step": 3070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007094566744061769,
+      "loss": 2.9025,
+      "step": 3080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007091116918268401,
+      "loss": 2.9126,
+      "step": 3090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007087667092475033,
+      "loss": 2.908,
+      "step": 3100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007084217266681664,
+      "loss": 2.8972,
+      "step": 3110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007080767440888296,
+      "loss": 2.8977,
+      "step": 3120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007077317615094928,
+      "loss": 2.8933,
+      "step": 3130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000707386778930156,
+      "loss": 2.9034,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007070417963508192,
+      "loss": 2.8959,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007066968137714823,
+      "loss": 2.9185,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007063518311921455,
+      "loss": 2.9064,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007060068486128087,
+      "loss": 2.9095,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007056618660334718,
+      "loss": 2.8941,
+      "step": 3190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000705316883454135,
+      "loss": 2.9054,
+      "step": 3200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007049719008747982,
+      "loss": 2.9048,
+      "step": 3210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007046269182954613,
+      "loss": 2.908,
+      "step": 3220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007042819357161246,
+      "loss": 2.9036,
+      "step": 3230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007039369531367878,
+      "loss": 2.9055,
+      "step": 3240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007035919705574509,
+      "loss": 2.8964,
+      "step": 3250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007032469879781141,
+      "loss": 2.9128,
+      "step": 3260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007029020053987773,
+      "loss": 2.9075,
+      "step": 3270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007025570228194405,
+      "loss": 2.9069,
+      "step": 3280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007022120402401036,
+      "loss": 2.8996,
+      "step": 3290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007018670576607668,
+      "loss": 2.8968,
+      "step": 3300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00070152207508143,
+      "loss": 2.8955,
+      "step": 3310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007011770925020931,
+      "loss": 2.9113,
+      "step": 3320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007008321099227563,
+      "loss": 2.9123,
+      "step": 3330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007004871273434195,
+      "loss": 2.899,
+      "step": 3340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007001421447640826,
+      "loss": 2.8995,
+      "step": 3350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0006997971621847458,
+      "loss": 2.9076,
+      "step": 3360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000699452179605409,
+      "loss": 2.8947,
+      "step": 3370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0006991071970260721,
+      "loss": 2.9133,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0006987622144467353,
+      "loss": 2.9136,
+      "step": 3390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0006984172318673986,
+      "loss": 2.9039,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0006980722492880618,
+      "loss": 2.9008,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0006977272667087249,
+      "loss": 2.897,
+      "step": 3420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006973822841293881,
+      "loss": 2.8756,
+      "step": 3430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006970373015500513,
+      "loss": 2.9163,
+      "step": 3440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006966923189707144,
+      "loss": 2.8989,
+      "step": 3450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006963473363913776,
+      "loss": 2.9025,
+      "step": 3460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006960023538120408,
+      "loss": 2.8996,
+      "step": 3470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006956573712327039,
+      "loss": 2.9048,
+      "step": 3480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006953123886533671,
+      "loss": 2.9029,
+      "step": 3490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006949674060740303,
+      "loss": 2.9088,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006946224234946934,
+      "loss": 2.9117,
+      "step": 3510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006942774409153566,
+      "loss": 2.9224,
+      "step": 3520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006939324583360198,
+      "loss": 2.8946,
+      "step": 3530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000693587475756683,
+      "loss": 2.9058,
+      "step": 3540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006932424931773461,
+      "loss": 2.9024,
+      "step": 3550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006928975105980093,
+      "loss": 2.9153,
+      "step": 3560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006925525280186725,
+      "loss": 2.9138,
+      "step": 3570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006922075454393356,
+      "loss": 2.8857,
+      "step": 3580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006918625628599989,
+      "loss": 2.9096,
+      "step": 3590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006915175802806621,
+      "loss": 2.8862,
+      "step": 3600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006911725977013253,
+      "loss": 2.9092,
+      "step": 3610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006908276151219884,
+      "loss": 2.8989,
+      "step": 3620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006904826325426516,
+      "loss": 2.8952,
+      "step": 3630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006901376499633148,
+      "loss": 2.9227,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000689792667383978,
+      "loss": 2.9,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0006894476848046412,
+      "loss": 2.9015,
+      "step": 3660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006891027022253043,
+      "loss": 2.8951,
+      "step": 3670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006887577196459675,
+      "loss": 2.9099,
+      "step": 3680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006884127370666307,
+      "loss": 2.9065,
+      "step": 3690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006880677544872939,
+      "loss": 2.8996,
+      "step": 3700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000687722771907957,
+      "loss": 2.9044,
+      "step": 3710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006873777893286202,
+      "loss": 2.9041,
+      "step": 3720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006870328067492834,
+      "loss": 2.8965,
+      "step": 3730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006866878241699465,
+      "loss": 2.897,
+      "step": 3740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006863428415906098,
+      "loss": 2.9077,
+      "step": 3750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000685997859011273,
+      "loss": 2.9194,
+      "step": 3760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006856528764319361,
+      "loss": 2.9145,
+      "step": 3770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006853078938525993,
+      "loss": 2.8947,
+      "step": 3780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006849629112732625,
+      "loss": 2.9055,
+      "step": 3790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006846179286939257,
+      "loss": 2.9027,
+      "step": 3800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006842729461145888,
+      "loss": 2.9038,
+      "step": 3810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000683927963535252,
+      "loss": 2.9105,
+      "step": 3820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006835829809559152,
+      "loss": 2.8994,
+      "step": 3830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006832379983765783,
+      "loss": 2.9036,
+      "step": 3840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006828930157972415,
+      "loss": 2.9008,
+      "step": 3850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006825480332179047,
+      "loss": 2.8963,
+      "step": 3860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006822030506385678,
+      "loss": 2.9135,
+      "step": 3870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000681858068059231,
+      "loss": 2.9112,
+      "step": 3880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006815130854798942,
+      "loss": 2.8977,
+      "step": 3890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0006811681029005573,
+      "loss": 2.8974,
+      "step": 3900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006808231203212205,
+      "loss": 2.8886,
+      "step": 3910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006804781377418837,
+      "loss": 2.8985,
+      "step": 3920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000680133155162547,
+      "loss": 2.8999,
+      "step": 3930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006797881725832101,
+      "loss": 2.902,
+      "step": 3940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006794431900038733,
+      "loss": 2.9081,
+      "step": 3950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006790982074245365,
+      "loss": 2.8937,
+      "step": 3960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006787532248451996,
+      "loss": 2.8959,
+      "step": 3970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006784082422658628,
+      "loss": 2.9049,
+      "step": 3980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000678063259686526,
+      "loss": 2.9007,
+      "step": 3990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006777182771071891,
+      "loss": 2.9109,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006773732945278523,
+      "loss": 2.9129,
+      "step": 4010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006770283119485155,
+      "loss": 2.9111,
+      "step": 4020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006766833293691786,
+      "loss": 2.9026,
+      "step": 4030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006763383467898418,
+      "loss": 2.8984,
+      "step": 4040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000675993364210505,
+      "loss": 2.9026,
+      "step": 4050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006756483816311682,
+      "loss": 2.9092,
+      "step": 4060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006753033990518313,
+      "loss": 2.9081,
+      "step": 4070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006749584164724945,
+      "loss": 2.8953,
+      "step": 4080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006746134338931577,
+      "loss": 2.9071,
+      "step": 4090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006742684513138208,
+      "loss": 2.9056,
+      "step": 4100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000673923468734484,
+      "loss": 2.8903,
+      "step": 4110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006735784861551473,
+      "loss": 2.8997,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006732335035758104,
+      "loss": 2.9165,
+      "step": 4130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006728885209964736,
+      "loss": 2.8991,
+      "step": 4140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006725435384171368,
+      "loss": 2.9126,
+      "step": 4150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006721985558378,
+      "loss": 2.9076,
+      "step": 4160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006718535732584632,
+      "loss": 2.9094,
+      "step": 4170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006715085906791263,
+      "loss": 2.8852,
+      "step": 4180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006711636080997895,
+      "loss": 2.8832,
+      "step": 4190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006708186255204527,
+      "loss": 2.9025,
+      "step": 4200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006704736429411159,
+      "loss": 2.8955,
+      "step": 4210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006701286603617791,
+      "loss": 2.893,
+      "step": 4220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006697836777824422,
+      "loss": 2.9118,
+      "step": 4230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006694386952031054,
+      "loss": 2.909,
+      "step": 4240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006690937126237686,
+      "loss": 2.9177,
+      "step": 4250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006687487300444317,
+      "loss": 2.8967,
+      "step": 4260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000668403747465095,
+      "loss": 2.9072,
+      "step": 4270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006680587648857582,
+      "loss": 2.9054,
+      "step": 4280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006677137823064213,
+      "loss": 2.9094,
+      "step": 4290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006673687997270845,
+      "loss": 2.9005,
+      "step": 4300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006670238171477477,
+      "loss": 2.9057,
+      "step": 4310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006666788345684109,
+      "loss": 2.9001,
+      "step": 4320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000666333851989074,
+      "loss": 2.9143,
+      "step": 4330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006659888694097372,
+      "loss": 2.8965,
+      "step": 4340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006656438868304004,
+      "loss": 2.9029,
+      "step": 4350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006652989042510635,
+      "loss": 2.9011,
+      "step": 4360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006649539216717267,
+      "loss": 2.9013,
+      "step": 4370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006646089390923899,
+      "loss": 2.9078,
+      "step": 4380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000664263956513053,
+      "loss": 2.9095,
+      "step": 4390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006639189739337162,
+      "loss": 2.901,
+      "step": 4400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006635739913543794,
+      "loss": 2.9103,
+      "step": 4410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006632290087750425,
+      "loss": 2.9003,
+      "step": 4420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006628840261957057,
+      "loss": 2.9015,
+      "step": 4430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000662539043616369,
+      "loss": 2.8987,
+      "step": 4440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006621940610370322,
+      "loss": 2.9064,
+      "step": 4450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006618490784576953,
+      "loss": 2.8974,
+      "step": 4460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006615040958783585,
+      "loss": 2.8986,
+      "step": 4470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006611591132990217,
+      "loss": 2.9128,
+      "step": 4480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006608141307196848,
+      "loss": 2.9119,
+      "step": 4490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000660469148140348,
+      "loss": 2.9152,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006601241655610112,
+      "loss": 2.8938,
+      "step": 4510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006597791829816743,
+      "loss": 2.9044,
+      "step": 4520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006594342004023375,
+      "loss": 2.9159,
+      "step": 4530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006590892178230007,
+      "loss": 2.8953,
+      "step": 4540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006587442352436638,
+      "loss": 2.9027,
+      "step": 4550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000658399252664327,
+      "loss": 2.8906,
+      "step": 4560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006580542700849902,
+      "loss": 2.8989,
+      "step": 4570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006577092875056534,
+      "loss": 2.8934,
+      "step": 4580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006573643049263165,
+      "loss": 2.9016,
+      "step": 4590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006570193223469797,
+      "loss": 2.9079,
+      "step": 4600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006566743397676429,
+      "loss": 2.8986,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000656329357188306,
+      "loss": 2.9099,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006559843746089693,
+      "loss": 2.9049,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006556393920296325,
+      "loss": 2.9071,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006552944094502956,
+      "loss": 2.8946,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006549494268709588,
+      "loss": 2.9004,
+      "step": 4660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000654604444291622,
+      "loss": 2.9036,
+      "step": 4670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006542594617122852,
+      "loss": 2.9162,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006539144791329483,
+      "loss": 2.8987,
+      "step": 4690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006535694965536115,
+      "loss": 2.8994,
+      "step": 4700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006532245139742747,
+      "loss": 2.9175,
+      "step": 4710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006528795313949379,
+      "loss": 2.9043,
+      "step": 4720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006525345488156011,
+      "loss": 2.9072,
+      "step": 4730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006521895662362642,
+      "loss": 2.9012,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006518445836569274,
+      "loss": 2.9052,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006514996010775906,
+      "loss": 2.9051,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006511546184982538,
+      "loss": 2.897,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006508096359189169,
+      "loss": 2.9193,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006504646533395801,
+      "loss": 2.9009,
+      "step": 4790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006501196707602434,
+      "loss": 2.9101,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006497746881809065,
+      "loss": 2.9085,
+      "step": 4810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006494297056015697,
+      "loss": 2.9008,
+      "step": 4820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006490847230222329,
+      "loss": 2.8975,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006487397404428961,
+      "loss": 2.9077,
+      "step": 4840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006483947578635592,
+      "loss": 2.9099,
+      "step": 4850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006480497752842224,
+      "loss": 2.9065,
+      "step": 4860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006477047927048856,
+      "loss": 2.9129,
+      "step": 4870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006473598101255487,
+      "loss": 2.912,
+      "step": 4880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006470148275462119,
+      "loss": 2.8972,
+      "step": 4890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006466698449668751,
+      "loss": 2.8918,
+      "step": 4900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006463248623875382,
+      "loss": 2.9046,
+      "step": 4910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006459798798082014,
+      "loss": 2.9139,
+      "step": 4920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006456348972288646,
+      "loss": 2.915,
+      "step": 4930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006452899146495277,
+      "loss": 2.902,
+      "step": 4940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006449449320701909,
+      "loss": 2.9018,
+      "step": 4950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006445999494908541,
+      "loss": 2.8996,
+      "step": 4960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006442549669115174,
+      "loss": 2.8922,
+      "step": 4970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006439099843321805,
+      "loss": 2.9047,
+      "step": 4980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006435650017528437,
+      "loss": 2.9081,
+      "step": 4990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006432200191735069,
+      "loss": 2.9027,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00064287503659417,
+      "loss": 2.8941,
+      "step": 5010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006425300540148332,
+      "loss": 2.8949,
+      "step": 5020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006421850714354964,
+      "loss": 2.9031,
+      "step": 5030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006418400888561595,
+      "loss": 2.8952,
+      "step": 5040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006414951062768227,
+      "loss": 2.9016,
+      "step": 5050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006411501236974859,
+      "loss": 2.907,
+      "step": 5060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000640805141118149,
+      "loss": 2.8966,
+      "step": 5070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006404601585388122,
+      "loss": 2.9058,
+      "step": 5080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006401151759594754,
+      "loss": 2.9226,
+      "step": 5090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006397701933801386,
+      "loss": 2.9025,
+      "step": 5100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006394252108008017,
+      "loss": 2.8971,
+      "step": 5110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006390802282214649,
+      "loss": 2.9015,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006387352456421281,
+      "loss": 2.8988,
+      "step": 5130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006383902630627912,
+      "loss": 2.9156,
+      "step": 5140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006380452804834544,
+      "loss": 2.8986,
+      "step": 5150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006377002979041177,
+      "loss": 2.9008,
+      "step": 5160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006373553153247808,
+      "loss": 2.9034,
+      "step": 5170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000637010332745444,
+      "loss": 2.9091,
+      "step": 5180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006366653501661072,
+      "loss": 2.9212,
+      "step": 5190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006363203675867703,
+      "loss": 2.9184,
+      "step": 5200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006359753850074335,
+      "loss": 2.8988,
+      "step": 5210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006356304024280967,
+      "loss": 2.912,
+      "step": 5220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006352854198487599,
+      "loss": 2.8937,
+      "step": 5230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006349404372694231,
+      "loss": 2.891,
+      "step": 5240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006345954546900862,
+      "loss": 2.9009,
+      "step": 5250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006342504721107494,
+      "loss": 2.9075,
+      "step": 5260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006339054895314126,
+      "loss": 2.905,
+      "step": 5270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006335605069520758,
+      "loss": 2.9103,
+      "step": 5280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006332155243727389,
+      "loss": 2.9017,
+      "step": 5290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006328705417934021,
+      "loss": 2.8917,
+      "step": 5300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006325255592140653,
+      "loss": 2.9085,
+      "step": 5310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006321805766347286,
+      "loss": 2.9049,
+      "step": 5320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006318355940553917,
+      "loss": 2.9008,
+      "step": 5330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006314906114760549,
+      "loss": 2.8997,
+      "step": 5340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006311456288967181,
+      "loss": 2.8975,
+      "step": 5350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006308006463173813,
+      "loss": 2.9058,
+      "step": 5360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006304556637380444,
+      "loss": 2.9058,
+      "step": 5370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006301106811587076,
+      "loss": 2.9139,
+      "step": 5380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006297656985793708,
+      "loss": 2.8928,
+      "step": 5390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006294207160000339,
+      "loss": 2.8993,
+      "step": 5400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006290757334206971,
+      "loss": 2.9112,
+      "step": 5410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006287307508413603,
+      "loss": 2.903,
+      "step": 5420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006283857682620234,
+      "loss": 2.9091,
+      "step": 5430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006280407856826866,
+      "loss": 2.9,
+      "step": 5440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006276958031033498,
+      "loss": 2.9024,
+      "step": 5450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006273508205240129,
+      "loss": 2.9201,
+      "step": 5460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006270058379446761,
+      "loss": 2.9093,
+      "step": 5470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006266608553653393,
+      "loss": 2.9098,
+      "step": 5480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006263158727860026,
+      "loss": 2.9064,
+      "step": 5490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006259708902066657,
+      "loss": 2.9041,
+      "step": 5500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006256259076273289,
+      "loss": 2.8972,
+      "step": 5510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006252809250479921,
+      "loss": 2.8998,
+      "step": 5520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006249359424686552,
+      "loss": 2.899,
+      "step": 5530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006245909598893184,
+      "loss": 2.9001,
+      "step": 5540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006242459773099816,
+      "loss": 2.9031,
+      "step": 5550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006239009947306447,
+      "loss": 2.8887,
+      "step": 5560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006235560121513079,
+      "loss": 2.9057,
+      "step": 5570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006232110295719711,
+      "loss": 2.9112,
+      "step": 5580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006228660469926342,
+      "loss": 2.9055,
+      "step": 5590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006225210644132974,
+      "loss": 2.8992,
+      "step": 5600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006221760818339606,
+      "loss": 2.9092,
+      "step": 5610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006218310992546238,
+      "loss": 2.9104,
+      "step": 5620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006214861166752869,
+      "loss": 2.8858,
+      "step": 5630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006211411340959501,
+      "loss": 2.9018,
+      "step": 5640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006207961515166133,
+      "loss": 2.9045,
+      "step": 5650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006204511689372764,
+      "loss": 2.8924,
+      "step": 5660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006201061863579396,
+      "loss": 2.8978,
+      "step": 5670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006197612037786029,
+      "loss": 2.9178,
+      "step": 5680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000619416221199266,
+      "loss": 2.9096,
+      "step": 5690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006190712386199292,
+      "loss": 2.9025,
+      "step": 5700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006187262560405924,
+      "loss": 2.9081,
+      "step": 5710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006183812734612555,
+      "loss": 2.9091,
+      "step": 5720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006180362908819187,
+      "loss": 2.9043,
+      "step": 5730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006176913083025819,
+      "loss": 2.9025,
+      "step": 5740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006173463257232451,
+      "loss": 2.9025,
+      "step": 5750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006170013431439082,
+      "loss": 2.9266,
+      "step": 5760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006166563605645714,
+      "loss": 2.9148,
+      "step": 5770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006163113779852346,
+      "loss": 2.9026,
+      "step": 5780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0006159663954058978,
+      "loss": 2.9026,
+      "step": 5790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006156214128265609,
+      "loss": 2.9023,
+      "step": 5800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006152764302472241,
+      "loss": 2.8919,
+      "step": 5810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006149314476678873,
+      "loss": 2.9054,
+      "step": 5820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006145864650885505,
+      "loss": 2.9,
+      "step": 5830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006142414825092138,
+      "loss": 2.9112,
+      "step": 5840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006138964999298769,
+      "loss": 2.8906,
+      "step": 5850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006135515173505401,
+      "loss": 2.9089,
+      "step": 5860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006132065347712033,
+      "loss": 2.9167,
+      "step": 5870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006128615521918665,
+      "loss": 2.8996,
+      "step": 5880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006125165696125296,
+      "loss": 2.9079,
+      "step": 5890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006121715870331928,
+      "loss": 2.8949,
+      "step": 5900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000611826604453856,
+      "loss": 2.9097,
+      "step": 5910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006114816218745191,
+      "loss": 2.9114,
+      "step": 5920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006111366392951823,
+      "loss": 2.9106,
+      "step": 5930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006107916567158455,
+      "loss": 2.9108,
+      "step": 5940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006104466741365086,
+      "loss": 2.9008,
+      "step": 5950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006101016915571718,
+      "loss": 2.9008,
+      "step": 5960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000609756708977835,
+      "loss": 2.9202,
+      "step": 5970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006094117263984981,
+      "loss": 2.8971,
+      "step": 5980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006090667438191613,
+      "loss": 2.9103,
+      "step": 5990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006087217612398245,
+      "loss": 2.9041,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006083767786604877,
+      "loss": 2.9048,
+      "step": 6010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0006080317960811508,
+      "loss": 2.9055,
+      "step": 6020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006076868135018141,
+      "loss": 2.9015,
+      "step": 6030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006073418309224773,
+      "loss": 2.9019,
+      "step": 6040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006069968483431404,
+      "loss": 2.9024,
+      "step": 6050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006066518657638036,
+      "loss": 2.9055,
+      "step": 6060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006063068831844668,
+      "loss": 2.9112,
+      "step": 6070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006059619006051299,
+      "loss": 2.9278,
+      "step": 6080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006056169180257931,
+      "loss": 2.8932,
+      "step": 6090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006052719354464563,
+      "loss": 2.9016,
+      "step": 6100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006049269528671194,
+      "loss": 2.8925,
+      "step": 6110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006045819702877826,
+      "loss": 2.8834,
+      "step": 6120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006042369877084458,
+      "loss": 2.9105,
+      "step": 6130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000603892005129109,
+      "loss": 2.9172,
+      "step": 6140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006035470225497721,
+      "loss": 2.9143,
+      "step": 6150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006032020399704353,
+      "loss": 2.9002,
+      "step": 6160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006028570573910985,
+      "loss": 2.9001,
+      "step": 6170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006025120748117616,
+      "loss": 2.9041,
+      "step": 6180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006021670922324248,
+      "loss": 2.9136,
+      "step": 6190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000601822109653088,
+      "loss": 2.8989,
+      "step": 6200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006014771270737512,
+      "loss": 2.9045,
+      "step": 6210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006011321444944144,
+      "loss": 2.9029,
+      "step": 6220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006007871619150776,
+      "loss": 2.9112,
+      "step": 6230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006004421793357407,
+      "loss": 2.8995,
+      "step": 6240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0006000971967564039,
+      "loss": 2.9056,
+      "step": 6250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005997522141770671,
+      "loss": 2.9104,
+      "step": 6260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005994072315977303,
+      "loss": 2.9068,
+      "step": 6270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005990622490183934,
+      "loss": 2.9214,
+      "step": 6280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005987172664390566,
+      "loss": 2.9106,
+      "step": 6290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005983722838597198,
+      "loss": 2.8999,
+      "step": 6300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005980273012803829,
+      "loss": 2.9008,
+      "step": 6310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005976823187010461,
+      "loss": 2.91,
+      "step": 6320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005973373361217093,
+      "loss": 2.9002,
+      "step": 6330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005969923535423725,
+      "loss": 2.8994,
+      "step": 6340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005966473709630357,
+      "loss": 2.9054,
+      "step": 6350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005963023883836988,
+      "loss": 2.9069,
+      "step": 6360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000595957405804362,
+      "loss": 2.8992,
+      "step": 6370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005956124232250253,
+      "loss": 2.8997,
+      "step": 6380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005952674406456885,
+      "loss": 2.8925,
+      "step": 6390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005949224580663517,
+      "loss": 2.894,
+      "step": 6400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005945774754870148,
+      "loss": 2.9215,
+      "step": 6410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000594232492907678,
+      "loss": 2.9137,
+      "step": 6420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005938875103283412,
+      "loss": 2.9014,
+      "step": 6430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005935425277490043,
+      "loss": 2.9065,
+      "step": 6440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005931975451696675,
+      "loss": 2.9107,
+      "step": 6450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005928525625903307,
+      "loss": 2.9045,
+      "step": 6460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005925075800109938,
+      "loss": 2.909,
+      "step": 6470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000592162597431657,
+      "loss": 2.9004,
+      "step": 6480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005918176148523202,
+      "loss": 2.9077,
+      "step": 6490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005914726322729833,
+      "loss": 2.8983,
+      "step": 6500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005911276496936465,
+      "loss": 2.9071,
+      "step": 6510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005907826671143097,
+      "loss": 2.9025,
+      "step": 6520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000590437684534973,
+      "loss": 2.9097,
+      "step": 6530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000590092701955636,
+      "loss": 2.9068,
+      "step": 6540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005897477193762993,
+      "loss": 2.9074,
+      "step": 6550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005894027367969625,
+      "loss": 2.8998,
+      "step": 6560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005890577542176256,
+      "loss": 2.9003,
+      "step": 6570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005887127716382888,
+      "loss": 2.9116,
+      "step": 6580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000588367789058952,
+      "loss": 2.9078,
+      "step": 6590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005880228064796151,
+      "loss": 2.9022,
+      "step": 6600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005876778239002783,
+      "loss": 2.8916,
+      "step": 6610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005873328413209415,
+      "loss": 2.9104,
+      "step": 6620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005869878587416046,
+      "loss": 2.9072,
+      "step": 6630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005866428761622678,
+      "loss": 2.9163,
+      "step": 6640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000586297893582931,
+      "loss": 2.9062,
+      "step": 6650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005859529110035942,
+      "loss": 2.8917,
+      "step": 6660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005856079284242573,
+      "loss": 2.9044,
+      "step": 6670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005852629458449205,
+      "loss": 2.9225,
+      "step": 6680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005849179632655837,
+      "loss": 2.8951,
+      "step": 6690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005845729806862468,
+      "loss": 2.9004,
+      "step": 6700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00058422799810691,
+      "loss": 2.8982,
+      "step": 6710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005838830155275733,
+      "loss": 2.9175,
+      "step": 6720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005835380329482364,
+      "loss": 2.9122,
+      "step": 6730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005831930503688996,
+      "loss": 2.9009,
+      "step": 6740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005828480677895628,
+      "loss": 2.8928,
+      "step": 6750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005825030852102259,
+      "loss": 2.8991,
+      "step": 6760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005821581026308891,
+      "loss": 2.916,
+      "step": 6770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005818131200515523,
+      "loss": 2.9041,
+      "step": 6780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005814681374722155,
+      "loss": 2.9251,
+      "step": 6790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005811231548928786,
+      "loss": 2.9006,
+      "step": 6800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005807781723135418,
+      "loss": 2.8918,
+      "step": 6810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000580433189734205,
+      "loss": 2.9042,
+      "step": 6820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005800882071548681,
+      "loss": 2.9007,
+      "step": 6830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005797432245755313,
+      "loss": 2.8986,
+      "step": 6840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005793982419961945,
+      "loss": 2.9052,
+      "step": 6850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005790532594168577,
+      "loss": 2.9059,
+      "step": 6860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005787082768375208,
+      "loss": 2.9048,
+      "step": 6870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000578363294258184,
+      "loss": 2.9066,
+      "step": 6880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005780183116788472,
+      "loss": 2.8862,
+      "step": 6890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005776733290995105,
+      "loss": 2.8996,
+      "step": 6900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005773283465201737,
+      "loss": 2.896,
+      "step": 6910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005769833639408368,
+      "loss": 2.9101,
+      "step": 6920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005766383813615,
+      "loss": 2.8966,
+      "step": 6930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005762933987821632,
+      "loss": 2.8963,
+      "step": 6940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005759484162028264,
+      "loss": 2.8939,
+      "step": 6950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005756034336234895,
+      "loss": 2.906,
+      "step": 6960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005752584510441527,
+      "loss": 2.9179,
+      "step": 6970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005749134684648159,
+      "loss": 2.8967,
+      "step": 6980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000574568485885479,
+      "loss": 2.8917,
+      "step": 6990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005742235033061422,
+      "loss": 2.906,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005738785207268054,
+      "loss": 2.9063,
+      "step": 7010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005735335381474685,
+      "loss": 2.9098,
+      "step": 7020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005731885555681317,
+      "loss": 2.9044,
+      "step": 7030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005728435729887949,
+      "loss": 2.9039,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005724985904094581,
+      "loss": 2.9051,
+      "step": 7050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005721536078301212,
+      "loss": 2.8998,
+      "step": 7060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005718086252507845,
+      "loss": 2.9119,
+      "step": 7070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005714636426714477,
+      "loss": 2.9053,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005711186600921108,
+      "loss": 2.8991,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000570773677512774,
+      "loss": 2.8762,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005704286949334372,
+      "loss": 2.8988,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005700837123541003,
+      "loss": 2.8914,
+      "step": 7120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005697387297747635,
+      "loss": 2.9083,
+      "step": 7130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005693937471954267,
+      "loss": 2.9062,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005690487646160898,
+      "loss": 2.9059,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000568703782036753,
+      "loss": 2.905,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005683587994574162,
+      "loss": 2.8922,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005680138168780794,
+      "loss": 2.9034,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005676688342987425,
+      "loss": 2.9107,
+      "step": 7190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005673238517194057,
+      "loss": 2.8958,
+      "step": 7200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005669788691400689,
+      "loss": 2.9103,
+      "step": 7210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000566633886560732,
+      "loss": 2.9115,
+      "step": 7220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005662889039813952,
+      "loss": 2.9001,
+      "step": 7230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005659439214020585,
+      "loss": 2.9042,
+      "step": 7240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005655989388227216,
+      "loss": 2.8986,
+      "step": 7250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005652539562433848,
+      "loss": 2.918,
+      "step": 7260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000564908973664048,
+      "loss": 2.9032,
+      "step": 7270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005645639910847111,
+      "loss": 2.908,
+      "step": 7280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005642190085053743,
+      "loss": 2.8995,
+      "step": 7290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005638740259260375,
+      "loss": 2.8984,
+      "step": 7300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005635290433467007,
+      "loss": 2.9031,
+      "step": 7310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005631840607673638,
+      "loss": 2.8951,
+      "step": 7320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000562839078188027,
+      "loss": 2.903,
+      "step": 7330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005624940956086902,
+      "loss": 2.9079,
+      "step": 7340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005621491130293533,
+      "loss": 2.9078,
+      "step": 7350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005618041304500165,
+      "loss": 2.9031,
+      "step": 7360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005614591478706797,
+      "loss": 2.9108,
+      "step": 7370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005611141652913428,
+      "loss": 2.9265,
+      "step": 7380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000560769182712006,
+      "loss": 2.9104,
+      "step": 7390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005604242001326692,
+      "loss": 2.8965,
+      "step": 7400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005600792175533324,
+      "loss": 2.8995,
+      "step": 7410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005597342349739957,
+      "loss": 2.9088,
+      "step": 7420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005593892523946588,
+      "loss": 2.9053,
+      "step": 7430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000559044269815322,
+      "loss": 2.9089,
+      "step": 7440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005586992872359852,
+      "loss": 2.894,
+      "step": 7450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005583543046566484,
+      "loss": 2.9076,
+      "step": 7460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005580093220773116,
+      "loss": 2.9031,
+      "step": 7470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005576643394979747,
+      "loss": 2.8918,
+      "step": 7480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005573193569186379,
+      "loss": 2.9068,
+      "step": 7490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005569743743393011,
+      "loss": 2.9063,
+      "step": 7500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005566293917599642,
+      "loss": 2.9068,
+      "step": 7510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005562844091806274,
+      "loss": 2.9021,
+      "step": 7520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005559394266012906,
+      "loss": 2.9001,
+      "step": 7530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005555944440219537,
+      "loss": 2.9046,
+      "step": 7540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005552494614426169,
+      "loss": 2.8998,
+      "step": 7550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005549044788632801,
+      "loss": 2.9014,
+      "step": 7560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005545594962839433,
+      "loss": 2.9064,
+      "step": 7570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005542145137046064,
+      "loss": 2.8991,
+      "step": 7580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005538695311252697,
+      "loss": 2.9038,
+      "step": 7590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005535245485459329,
+      "loss": 2.9106,
+      "step": 7600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000553179565966596,
+      "loss": 2.912,
+      "step": 7610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005528345833872592,
+      "loss": 2.898,
+      "step": 7620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005524896008079224,
+      "loss": 2.9093,
+      "step": 7630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005521446182285855,
+      "loss": 2.9085,
+      "step": 7640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005517996356492487,
+      "loss": 2.891,
+      "step": 7650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005514546530699119,
+      "loss": 2.9049,
+      "step": 7660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000551109670490575,
+      "loss": 2.8957,
+      "step": 7670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005507646879112382,
+      "loss": 2.904,
+      "step": 7680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005504197053319014,
+      "loss": 2.8954,
+      "step": 7690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005500747227525646,
+      "loss": 2.9023,
+      "step": 7700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005497297401732277,
+      "loss": 2.8996,
+      "step": 7710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005493847575938909,
+      "loss": 2.9086,
+      "step": 7720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005490397750145541,
+      "loss": 2.9006,
+      "step": 7730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005486947924352172,
+      "loss": 2.907,
+      "step": 7740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005483498098558804,
+      "loss": 2.902,
+      "step": 7750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005480048272765436,
+      "loss": 2.899,
+      "step": 7760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005476598446972067,
+      "loss": 2.9029,
+      "step": 7770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00054731486211787,
+      "loss": 2.9087,
+      "step": 7780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005469698795385332,
+      "loss": 2.9089,
+      "step": 7790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005466248969591963,
+      "loss": 2.9045,
+      "step": 7800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005462799143798595,
+      "loss": 2.9084,
+      "step": 7810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005459349318005227,
+      "loss": 2.9097,
+      "step": 7820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005455899492211859,
+      "loss": 2.9127,
+      "step": 7830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000545244966641849,
+      "loss": 2.898,
+      "step": 7840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005448999840625122,
+      "loss": 2.9055,
+      "step": 7850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005445550014831754,
+      "loss": 2.8964,
+      "step": 7860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005442100189038385,
+      "loss": 2.8931,
+      "step": 7870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005438650363245017,
+      "loss": 2.9125,
+      "step": 7880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005435200537451649,
+      "loss": 2.905,
+      "step": 7890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000543175071165828,
+      "loss": 2.9029,
+      "step": 7900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005428300885864912,
+      "loss": 2.9123,
+      "step": 7910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005424851060071544,
+      "loss": 2.8984,
+      "step": 7920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005421401234278176,
+      "loss": 2.8984,
+      "step": 7930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005417951408484807,
+      "loss": 2.9112,
+      "step": 7940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000541450158269144,
+      "loss": 2.8959,
+      "step": 7950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005411051756898072,
+      "loss": 2.9235,
+      "step": 7960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005407601931104704,
+      "loss": 2.9008,
+      "step": 7970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005404152105311336,
+      "loss": 2.908,
+      "step": 7980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005400702279517967,
+      "loss": 2.9083,
+      "step": 7990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005397252453724599,
+      "loss": 2.9081,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005393802627931231,
+      "loss": 2.913,
+      "step": 8010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005390352802137863,
+      "loss": 2.8984,
+      "step": 8020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005386902976344494,
+      "loss": 2.9068,
+      "step": 8030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005383453150551126,
+      "loss": 2.8967,
+      "step": 8040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005380003324757758,
+      "loss": 2.897,
+      "step": 8050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005376553498964389,
+      "loss": 2.9023,
+      "step": 8060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005373103673171021,
+      "loss": 2.9036,
+      "step": 8070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005369653847377653,
+      "loss": 2.9039,
+      "step": 8080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005366204021584285,
+      "loss": 2.8926,
+      "step": 8090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005362754195790916,
+      "loss": 2.9042,
+      "step": 8100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005359304369997549,
+      "loss": 2.9103,
+      "step": 8110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005355854544204181,
+      "loss": 2.8953,
+      "step": 8120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005352404718410812,
+      "loss": 2.9095,
+      "step": 8130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005348954892617444,
+      "loss": 2.903,
+      "step": 8140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005345505066824076,
+      "loss": 2.9,
+      "step": 8150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005342055241030707,
+      "loss": 2.9095,
+      "step": 8160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005338605415237339,
+      "loss": 2.9034,
+      "step": 8170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005335155589443971,
+      "loss": 2.8965,
+      "step": 8180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005331705763650602,
+      "loss": 2.896,
+      "step": 8190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005328255937857234,
+      "loss": 2.8925,
+      "step": 8200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005324806112063866,
+      "loss": 2.8992,
+      "step": 8210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005321356286270498,
+      "loss": 2.8988,
+      "step": 8220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005317906460477129,
+      "loss": 2.9093,
+      "step": 8230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005314456634683761,
+      "loss": 2.9082,
+      "step": 8240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005311006808890393,
+      "loss": 2.9122,
+      "step": 8250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005307556983097024,
+      "loss": 2.8879,
+      "step": 8260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005304107157303656,
+      "loss": 2.9015,
+      "step": 8270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005300657331510288,
+      "loss": 2.9098,
+      "step": 8280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000529720750571692,
+      "loss": 2.91,
+      "step": 8290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005293757679923552,
+      "loss": 2.9075,
+      "step": 8300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005290307854130184,
+      "loss": 2.9047,
+      "step": 8310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005286858028336815,
+      "loss": 2.8967,
+      "step": 8320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005283408202543447,
+      "loss": 2.9011,
+      "step": 8330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005279958376750079,
+      "loss": 2.9109,
+      "step": 8340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005276508550956711,
+      "loss": 2.904,
+      "step": 8350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005273058725163342,
+      "loss": 2.9119,
+      "step": 8360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005269608899369974,
+      "loss": 2.9032,
+      "step": 8370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005266159073576606,
+      "loss": 2.9044,
+      "step": 8380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005262709247783237,
+      "loss": 2.9052,
+      "step": 8390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005259259421989869,
+      "loss": 2.9119,
+      "step": 8400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005255809596196501,
+      "loss": 2.9125,
+      "step": 8410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005252359770403132,
+      "loss": 2.8925,
+      "step": 8420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005248909944609764,
+      "loss": 2.9155,
+      "step": 8430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005245460118816396,
+      "loss": 2.9052,
+      "step": 8440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005242010293023027,
+      "loss": 2.888,
+      "step": 8450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005238560467229659,
+      "loss": 2.9058,
+      "step": 8460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005235110641436292,
+      "loss": 2.9007,
+      "step": 8470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005231660815642924,
+      "loss": 2.8932,
+      "step": 8480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005228210989849556,
+      "loss": 2.8937,
+      "step": 8490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005224761164056187,
+      "loss": 2.9206,
+      "step": 8500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005221311338262819,
+      "loss": 2.9089,
+      "step": 8510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005217861512469451,
+      "loss": 2.916,
+      "step": 8520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005214411686676083,
+      "loss": 2.9053,
+      "step": 8530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005210961860882715,
+      "loss": 2.8997,
+      "step": 8540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005207512035089346,
+      "loss": 2.8946,
+      "step": 8550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005204062209295978,
+      "loss": 2.9249,
+      "step": 8560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000520061238350261,
+      "loss": 2.8947,
+      "step": 8570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005197162557709241,
+      "loss": 2.897,
+      "step": 8580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005193712731915873,
+      "loss": 2.8989,
+      "step": 8590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005190262906122505,
+      "loss": 2.8937,
+      "step": 8600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005186813080329137,
+      "loss": 2.9022,
+      "step": 8610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005183363254535768,
+      "loss": 2.9171,
+      "step": 8620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00051799134287424,
+      "loss": 2.9054,
+      "step": 8630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005176463602949033,
+      "loss": 2.9095,
+      "step": 8640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005173013777155664,
+      "loss": 2.8993,
+      "step": 8650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005169563951362296,
+      "loss": 2.8891,
+      "step": 8660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005166114125568928,
+      "loss": 2.9088,
+      "step": 8670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005162664299775559,
+      "loss": 2.9196,
+      "step": 8680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005159214473982191,
+      "loss": 2.9137,
+      "step": 8690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005155764648188823,
+      "loss": 2.9003,
+      "step": 8700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005152314822395454,
+      "loss": 2.9065,
+      "step": 8710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005148864996602086,
+      "loss": 2.9077,
+      "step": 8720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005145415170808718,
+      "loss": 2.9094,
+      "step": 8730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000514196534501535,
+      "loss": 2.9062,
+      "step": 8740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005138515519221981,
+      "loss": 2.9215,
+      "step": 8750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005135065693428613,
+      "loss": 2.9042,
+      "step": 8760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005131615867635245,
+      "loss": 2.9086,
+      "step": 8770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005128166041841876,
+      "loss": 2.912,
+      "step": 8780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005124716216048508,
+      "loss": 2.9086,
+      "step": 8790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000512126639025514,
+      "loss": 2.902,
+      "step": 8800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005117816564461771,
+      "loss": 2.8941,
+      "step": 8810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005114366738668404,
+      "loss": 2.8883,
+      "step": 8820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005110916912875036,
+      "loss": 2.9077,
+      "step": 8830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005107467087081667,
+      "loss": 2.9086,
+      "step": 8840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005104017261288299,
+      "loss": 2.9143,
+      "step": 8850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005100567435494931,
+      "loss": 2.9144,
+      "step": 8860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005097117609701563,
+      "loss": 2.9068,
+      "step": 8870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005093667783908194,
+      "loss": 2.9075,
+      "step": 8880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005090217958114826,
+      "loss": 2.8986,
+      "step": 8890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005086768132321458,
+      "loss": 2.9131,
+      "step": 8900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005083318306528089,
+      "loss": 2.9015,
+      "step": 8910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005079868480734721,
+      "loss": 2.9067,
+      "step": 8920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005076418654941353,
+      "loss": 2.8943,
+      "step": 8930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005072968829147984,
+      "loss": 2.904,
+      "step": 8940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005069519003354616,
+      "loss": 2.9213,
+      "step": 8950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005066069177561248,
+      "loss": 2.9116,
+      "step": 8960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005062619351767879,
+      "loss": 2.908,
+      "step": 8970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005059169525974511,
+      "loss": 2.9008,
+      "step": 8980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005055719700181144,
+      "loss": 2.9139,
+      "step": 8990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005052269874387776,
+      "loss": 2.8939,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005048820048594407,
+      "loss": 2.8875,
+      "step": 9010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005045370222801039,
+      "loss": 2.9004,
+      "step": 9020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005041920397007671,
+      "loss": 2.8958,
+      "step": 9030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005038470571214303,
+      "loss": 2.9078,
+      "step": 9040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005035020745420935,
+      "loss": 2.9037,
+      "step": 9050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005031570919627566,
+      "loss": 2.9076,
+      "step": 9060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005028121093834198,
+      "loss": 2.9056,
+      "step": 9070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000502467126804083,
+      "loss": 2.8901,
+      "step": 9080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005021221442247462,
+      "loss": 2.9015,
+      "step": 9090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005017771616454093,
+      "loss": 2.9052,
+      "step": 9100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005014321790660725,
+      "loss": 2.9077,
+      "step": 9110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005010871964867357,
+      "loss": 2.9045,
+      "step": 9120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005007422139073989,
+      "loss": 2.8961,
+      "step": 9130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000500397231328062,
+      "loss": 2.89,
+      "step": 9140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005000522487487252,
+      "loss": 2.9013,
+      "step": 9150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004997072661693885,
+      "loss": 2.9046,
+      "step": 9160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004993622835900516,
+      "loss": 2.9032,
+      "step": 9170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004990173010107148,
+      "loss": 2.9021,
+      "step": 9180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000498672318431378,
+      "loss": 2.9147,
+      "step": 9190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004983273358520411,
+      "loss": 2.9011,
+      "step": 9200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004979823532727043,
+      "loss": 2.9028,
+      "step": 9210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004976373706933675,
+      "loss": 2.892,
+      "step": 9220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004972923881140306,
+      "loss": 2.9025,
+      "step": 9230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004969474055346938,
+      "loss": 2.89,
+      "step": 9240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000496602422955357,
+      "loss": 2.9169,
+      "step": 9250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004962574403760202,
+      "loss": 2.9126,
+      "step": 9260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004959124577966833,
+      "loss": 2.9011,
+      "step": 9270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004955674752173465,
+      "loss": 2.8993,
+      "step": 9280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004952224926380097,
+      "loss": 2.9103,
+      "step": 9290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004948775100586728,
+      "loss": 2.9172,
+      "step": 9300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000494532527479336,
+      "loss": 2.9037,
+      "step": 9310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004941875448999992,
+      "loss": 2.9075,
+      "step": 9320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004938425623206623,
+      "loss": 2.9051,
+      "step": 9330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004934975797413256,
+      "loss": 2.9087,
+      "step": 9340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004931525971619888,
+      "loss": 2.9196,
+      "step": 9350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004928076145826519,
+      "loss": 2.9098,
+      "step": 9360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004924626320033151,
+      "loss": 2.9231,
+      "step": 9370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004921176494239783,
+      "loss": 2.9151,
+      "step": 9380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004917726668446415,
+      "loss": 2.9018,
+      "step": 9390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004914276842653046,
+      "loss": 2.9143,
+      "step": 9400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004910827016859678,
+      "loss": 2.8958,
+      "step": 9410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000490737719106631,
+      "loss": 2.9088,
+      "step": 9420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004903927365272941,
+      "loss": 2.905,
+      "step": 9430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004900477539479573,
+      "loss": 2.9061,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004897027713686205,
+      "loss": 2.9071,
+      "step": 9450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004893577887892836,
+      "loss": 2.9052,
+      "step": 9460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004890128062099468,
+      "loss": 2.9071,
+      "step": 9470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00048866782363061,
+      "loss": 2.9154,
+      "step": 9480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004883228410512731,
+      "loss": 2.9061,
+      "step": 9490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004879778584719364,
+      "loss": 2.9053,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004876328758925996,
+      "loss": 2.8968,
+      "step": 9510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004872878933132628,
+      "loss": 2.8932,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004869429107339259,
+      "loss": 2.9021,
+      "step": 9530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004865979281545891,
+      "loss": 2.9061,
+      "step": 9540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004862529455752523,
+      "loss": 2.905,
+      "step": 9550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004859079629959154,
+      "loss": 2.9131,
+      "step": 9560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00048556298041657863,
+      "loss": 2.8997,
+      "step": 9570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048521799783724184,
+      "loss": 2.9173,
+      "step": 9580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048487301525790494,
+      "loss": 2.9038,
+      "step": 9590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048452803267856815,
+      "loss": 2.8952,
+      "step": 9600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048418305009923136,
+      "loss": 2.8908,
+      "step": 9610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048383806751989446,
+      "loss": 2.9029,
+      "step": 9620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048349308494055767,
+      "loss": 2.9052,
+      "step": 9630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004831481023612209,
+      "loss": 2.9097,
+      "step": 9640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004828031197818841,
+      "loss": 2.9033,
+      "step": 9650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004824581372025472,
+      "loss": 2.9022,
+      "step": 9660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004821131546232104,
+      "loss": 2.9015,
+      "step": 9670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004817681720438736,
+      "loss": 2.9144,
+      "step": 9680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004814231894645367,
+      "loss": 2.9092,
+      "step": 9690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004810782068851999,
+      "loss": 2.9028,
+      "step": 9700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004807332243058631,
+      "loss": 2.891,
+      "step": 9710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00048038824172652627,
+      "loss": 2.9116,
+      "step": 9720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004800432591471895,
+      "loss": 2.907,
+      "step": 9730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00047969827656785263,
+      "loss": 2.8991,
+      "step": 9740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004793532939885158,
+      "loss": 2.8884,
+      "step": 9750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000479008311409179,
+      "loss": 2.9067,
+      "step": 9760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004786633288298422,
+      "loss": 2.9058,
+      "step": 9770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004783183462505054,
+      "loss": 2.9058,
+      "step": 9780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004779733636711685,
+      "loss": 2.8971,
+      "step": 9790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004776283810918317,
+      "loss": 2.9017,
+      "step": 9800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004772833985124949,
+      "loss": 2.8957,
+      "step": 9810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000476938415933158,
+      "loss": 2.9068,
+      "step": 9820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047659343335382123,
+      "loss": 2.9139,
+      "step": 9830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047624845077448444,
+      "loss": 2.9114,
+      "step": 9840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047590346819514754,
+      "loss": 2.909,
+      "step": 9850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047555848561581075,
+      "loss": 2.9073,
+      "step": 9860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047521350303647396,
+      "loss": 2.8998,
+      "step": 9870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047486852045713706,
+      "loss": 2.9105,
+      "step": 9880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047452353787780027,
+      "loss": 2.9067,
+      "step": 9890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004741785552984635,
+      "loss": 2.8971,
+      "step": 9900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004738335727191267,
+      "loss": 2.9028,
+      "step": 9910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004734885901397898,
+      "loss": 2.8977,
+      "step": 9920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000473143607560453,
+      "loss": 2.9158,
+      "step": 9930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004727986249811162,
+      "loss": 2.8922,
+      "step": 9940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004724536424017793,
+      "loss": 2.8953,
+      "step": 9950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004721086598224425,
+      "loss": 2.9039,
+      "step": 9960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004717636772431057,
+      "loss": 2.9164,
+      "step": 9970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047141869466376887,
+      "loss": 2.9015,
+      "step": 9980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004710737120844321,
+      "loss": 2.905,
+      "step": 9990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00047072872950509523,
+      "loss": 2.8908,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004703837469257584,
+      "loss": 2.9014,
+      "step": 10010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004700387643464216,
+      "loss": 2.9106,
+      "step": 10020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004696937817670848,
+      "loss": 2.9007,
+      "step": 10030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000469348799187748,
+      "loss": 2.9122,
+      "step": 10040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004690038166084111,
+      "loss": 2.9045,
+      "step": 10050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004686588340290743,
+      "loss": 2.8936,
+      "step": 10060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004683138514497375,
+      "loss": 2.894,
+      "step": 10070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004679688688704006,
+      "loss": 2.9114,
+      "step": 10080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046762388629106383,
+      "loss": 2.891,
+      "step": 10090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046727890371172704,
+      "loss": 2.9035,
+      "step": 10100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046693392113239014,
+      "loss": 2.8906,
+      "step": 10110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046658893855305335,
+      "loss": 2.9,
+      "step": 10120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046624395597371656,
+      "loss": 2.899,
+      "step": 10130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046589897339437966,
+      "loss": 2.9031,
+      "step": 10140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046555399081504287,
+      "loss": 2.9123,
+      "step": 10150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004652090082357061,
+      "loss": 2.8996,
+      "step": 10160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004648640256563693,
+      "loss": 2.9018,
+      "step": 10170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004645190430770324,
+      "loss": 2.8978,
+      "step": 10180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004641740604976956,
+      "loss": 2.9098,
+      "step": 10190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004638290779183588,
+      "loss": 2.9158,
+      "step": 10200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004634840953390219,
+      "loss": 2.8974,
+      "step": 10210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004631391127596851,
+      "loss": 2.9031,
+      "step": 10220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004627941301803483,
+      "loss": 2.9104,
+      "step": 10230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046244914760101147,
+      "loss": 2.9059,
+      "step": 10240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004621041650216746,
+      "loss": 2.8943,
+      "step": 10250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00046175918244233783,
+      "loss": 2.9006,
+      "step": 10260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000461414199863001,
+      "loss": 2.9085,
+      "step": 10270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0004610692172836642,
+      "loss": 2.9016,
+      "step": 10280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004607242347043274,
+      "loss": 2.9025,
+      "step": 10290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00046037925212499055,
+      "loss": 2.9072,
+      "step": 10300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004600342695456537,
+      "loss": 2.9063,
+      "step": 10310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004596892869663169,
+      "loss": 2.9117,
+      "step": 10320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004593443043869801,
+      "loss": 2.8866,
+      "step": 10330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004589993218076432,
+      "loss": 2.9138,
+      "step": 10340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045865433922830643,
+      "loss": 2.9162,
+      "step": 10350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045830935664896964,
+      "loss": 2.8962,
+      "step": 10360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045796437406963274,
+      "loss": 2.8971,
+      "step": 10370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045761939149029595,
+      "loss": 2.9081,
+      "step": 10380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045727440891095916,
+      "loss": 2.9123,
+      "step": 10390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045692942633162226,
+      "loss": 2.9051,
+      "step": 10400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045658444375228546,
+      "loss": 2.9153,
+      "step": 10410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045623946117294867,
+      "loss": 2.9094,
+      "step": 10420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004558944785936119,
+      "loss": 2.8977,
+      "step": 10430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000455549496014275,
+      "loss": 2.9081,
+      "step": 10440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004552045134349382,
+      "loss": 2.9031,
+      "step": 10450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004548595308556014,
+      "loss": 2.8978,
+      "step": 10460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004545145482762645,
+      "loss": 2.9091,
+      "step": 10470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004541695656969277,
+      "loss": 2.9107,
+      "step": 10480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004538245831175909,
+      "loss": 2.9117,
+      "step": 10490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045347960053825407,
+      "loss": 2.9039,
+      "step": 10500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004531346179589172,
+      "loss": 2.9038,
+      "step": 10510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00045278963537958043,
+      "loss": 2.9077,
+      "step": 10520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004524446528002436,
+      "loss": 2.9038,
+      "step": 10530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004520996702209068,
+      "loss": 2.8992,
+      "step": 10540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00045175468764157,
+      "loss": 2.901,
+      "step": 10550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00045140970506223315,
+      "loss": 2.8907,
+      "step": 10560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004510647224828963,
+      "loss": 2.9119,
+      "step": 10570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004507197399035595,
+      "loss": 2.8977,
+      "step": 10580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004503747573242227,
+      "loss": 2.9045,
+      "step": 10590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004500297747448858,
+      "loss": 2.8983,
+      "step": 10600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044968479216554903,
+      "loss": 2.8948,
+      "step": 10610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044933980958621224,
+      "loss": 2.9032,
+      "step": 10620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044899482700687534,
+      "loss": 2.9089,
+      "step": 10630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044864984442753855,
+      "loss": 2.9128,
+      "step": 10640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044830486184820176,
+      "loss": 2.9016,
+      "step": 10650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044795987926886486,
+      "loss": 2.9021,
+      "step": 10660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044761489668952806,
+      "loss": 2.8914,
+      "step": 10670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00044726991411019127,
+      "loss": 2.9146,
+      "step": 10680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004469249315308545,
+      "loss": 2.9174,
+      "step": 10690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004465799489515176,
+      "loss": 2.9047,
+      "step": 10700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004462349663721808,
+      "loss": 2.8913,
+      "step": 10710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000445889983792844,
+      "loss": 2.8965,
+      "step": 10720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004455450012135071,
+      "loss": 2.9029,
+      "step": 10730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004452000186341703,
+      "loss": 2.894,
+      "step": 10740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004448550360548335,
+      "loss": 2.9038,
+      "step": 10750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004445100534754966,
+      "loss": 2.9065,
+      "step": 10760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004441650708961598,
+      "loss": 2.9078,
+      "step": 10770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00044382008831682303,
+      "loss": 2.9028,
+      "step": 10780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004434751057374862,
+      "loss": 2.9023,
+      "step": 10790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004431301231581494,
+      "loss": 2.8873,
+      "step": 10800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00044278514057881254,
+      "loss": 2.906,
+      "step": 10810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00044244015799947575,
+      "loss": 2.9075,
+      "step": 10820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004420951754201389,
+      "loss": 2.8962,
+      "step": 10830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004417501928408021,
+      "loss": 2.9123,
+      "step": 10840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004414052102614653,
+      "loss": 2.915,
+      "step": 10850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004410602276821284,
+      "loss": 2.8959,
+      "step": 10860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00044071524510279163,
+      "loss": 2.9156,
+      "step": 10870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00044037026252345484,
+      "loss": 2.9087,
+      "step": 10880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00044002527994411794,
+      "loss": 2.9031,
+      "step": 10890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043968029736478115,
+      "loss": 2.897,
+      "step": 10900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043933531478544435,
+      "loss": 2.9077,
+      "step": 10910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043899033220610745,
+      "loss": 2.9051,
+      "step": 10920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043864534962677066,
+      "loss": 2.8984,
+      "step": 10930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00043830036704743387,
+      "loss": 2.9073,
+      "step": 10940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004379553844680971,
+      "loss": 2.9071,
+      "step": 10950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004376104018887602,
+      "loss": 2.9073,
+      "step": 10960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004372654193094234,
+      "loss": 2.8995,
+      "step": 10970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004369204367300866,
+      "loss": 2.8882,
+      "step": 10980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004365754541507497,
+      "loss": 2.8884,
+      "step": 10990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004362304715714129,
+      "loss": 2.9116,
+      "step": 11000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004358854889920761,
+      "loss": 2.9023,
+      "step": 11010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004355405064127392,
+      "loss": 2.9005,
+      "step": 11020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004351955238334024,
+      "loss": 2.9089,
+      "step": 11030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043485054125406563,
+      "loss": 2.9077,
+      "step": 11040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004345055586747288,
+      "loss": 2.8999,
+      "step": 11050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000434160576095392,
+      "loss": 2.9083,
+      "step": 11060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043381559351605514,
+      "loss": 2.8939,
+      "step": 11070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043347061093671835,
+      "loss": 2.9091,
+      "step": 11080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004331256283573815,
+      "loss": 2.9118,
+      "step": 11090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004327806457780447,
+      "loss": 2.9172,
+      "step": 11100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004324356631987079,
+      "loss": 2.9024,
+      "step": 11110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000432090680619371,
+      "loss": 2.9059,
+      "step": 11120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043174569804003423,
+      "loss": 2.8927,
+      "step": 11130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043140071546069744,
+      "loss": 2.9121,
+      "step": 11140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043105573288136054,
+      "loss": 2.914,
+      "step": 11150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043071075030202375,
+      "loss": 2.9034,
+      "step": 11160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043036576772268695,
+      "loss": 2.9111,
+      "step": 11170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00043002078514335005,
+      "loss": 2.9066,
+      "step": 11180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042967580256401326,
+      "loss": 2.9114,
+      "step": 11190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00042933081998467647,
+      "loss": 2.898,
+      "step": 11200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004289858374053397,
+      "loss": 2.8984,
+      "step": 11210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0004286408548260028,
+      "loss": 2.9131,
+      "step": 11220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000428295872246666,
+      "loss": 2.8961,
+      "step": 11230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004279508896673292,
+      "loss": 2.9015,
+      "step": 11240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004276059070879923,
+      "loss": 2.9198,
+      "step": 11250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004272609245086555,
+      "loss": 2.8989,
+      "step": 11260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004269159419293187,
+      "loss": 2.9028,
+      "step": 11270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004265709593499818,
+      "loss": 2.8909,
+      "step": 11280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000426225976770645,
+      "loss": 2.8932,
+      "step": 11290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004258809941913082,
+      "loss": 2.9077,
+      "step": 11300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004255360116119714,
+      "loss": 2.9187,
+      "step": 11310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042519102903263453,
+      "loss": 2.9131,
+      "step": 11320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042484604645329774,
+      "loss": 2.9061,
+      "step": 11330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042450106387396095,
+      "loss": 2.9104,
+      "step": 11340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004241560812946241,
+      "loss": 2.9053,
+      "step": 11350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004238110987152873,
+      "loss": 2.9052,
+      "step": 11360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042346611613595047,
+      "loss": 2.9006,
+      "step": 11370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004231211335566136,
+      "loss": 2.8972,
+      "step": 11380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042277615097727683,
+      "loss": 2.9176,
+      "step": 11390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042243116839794004,
+      "loss": 2.8931,
+      "step": 11400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042208618581860314,
+      "loss": 2.8997,
+      "step": 11410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042174120323926634,
+      "loss": 2.907,
+      "step": 11420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042139622065992955,
+      "loss": 2.906,
+      "step": 11430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042105123808059265,
+      "loss": 2.9088,
+      "step": 11440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042070625550125586,
+      "loss": 2.8965,
+      "step": 11450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00042036127292191907,
+      "loss": 2.9097,
+      "step": 11460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004200162903425823,
+      "loss": 2.9112,
+      "step": 11470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004196713077632454,
+      "loss": 2.9069,
+      "step": 11480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004193263251839086,
+      "loss": 2.8907,
+      "step": 11490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004189813426045718,
+      "loss": 2.9074,
+      "step": 11500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004186363600252349,
+      "loss": 2.8942,
+      "step": 11510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004182913774458981,
+      "loss": 2.9048,
+      "step": 11520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004179463948665613,
+      "loss": 2.9223,
+      "step": 11530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004176014122872244,
+      "loss": 2.9023,
+      "step": 11540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004172564297078876,
+      "loss": 2.903,
+      "step": 11550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004169114471285508,
+      "loss": 2.8989,
+      "step": 11560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000416566464549214,
+      "loss": 2.9106,
+      "step": 11570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041622148196987713,
+      "loss": 2.9102,
+      "step": 11580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041587649939054034,
+      "loss": 2.897,
+      "step": 11590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041553151681120355,
+      "loss": 2.8918,
+      "step": 11600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004151865342318667,
+      "loss": 2.8932,
+      "step": 11610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004148415516525299,
+      "loss": 2.8949,
+      "step": 11620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041449656907319307,
+      "loss": 2.909,
+      "step": 11630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0004141515864938562,
+      "loss": 2.9098,
+      "step": 11640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041380660391451943,
+      "loss": 2.9009,
+      "step": 11650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041346162133518264,
+      "loss": 2.8963,
+      "step": 11660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041311663875584574,
+      "loss": 2.9102,
+      "step": 11670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041277165617650894,
+      "loss": 2.903,
+      "step": 11680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041242667359717215,
+      "loss": 2.9005,
+      "step": 11690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00041208169101783525,
+      "loss": 2.9086,
+      "step": 11700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00041173670843849846,
+      "loss": 2.8966,
+      "step": 11710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00041139172585916167,
+      "loss": 2.915,
+      "step": 11720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004110467432798249,
+      "loss": 2.9064,
+      "step": 11730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000410701760700488,
+      "loss": 2.9155,
+      "step": 11740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004103567781211512,
+      "loss": 2.8924,
+      "step": 11750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004100117955418144,
+      "loss": 2.9024,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004096668129624775,
+      "loss": 2.9083,
+      "step": 11770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004093218303831407,
+      "loss": 2.902,
+      "step": 11780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004089768478038039,
+      "loss": 2.9007,
+      "step": 11790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000408631865224467,
+      "loss": 2.8958,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004082868826451302,
+      "loss": 2.8967,
+      "step": 11810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004079419000657934,
+      "loss": 2.9076,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004075969174864565,
+      "loss": 2.8999,
+      "step": 11830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040725193490711973,
+      "loss": 2.8985,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040690695232778294,
+      "loss": 2.8996,
+      "step": 11850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040656196974844615,
+      "loss": 2.9003,
+      "step": 11860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004062169871691093,
+      "loss": 2.8949,
+      "step": 11870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040587200458977246,
+      "loss": 2.9059,
+      "step": 11880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040552702201043566,
+      "loss": 2.9006,
+      "step": 11890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0004051820394310988,
+      "loss": 2.9095,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000404837056851762,
+      "loss": 2.911,
+      "step": 11910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040449207427242524,
+      "loss": 2.9084,
+      "step": 11920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040414709169308833,
+      "loss": 2.9054,
+      "step": 11930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00040380210911375154,
+      "loss": 2.9067,
+      "step": 11940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00040345712653441475,
+      "loss": 2.9037,
+      "step": 11950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00040311214395507785,
+      "loss": 2.9056,
+      "step": 11960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00040276716137574106,
+      "loss": 2.914,
+      "step": 11970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00040242217879640427,
+      "loss": 2.912,
+      "step": 11980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004020771962170675,
+      "loss": 2.9176,
+      "step": 11990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004017322136377306,
+      "loss": 2.9026,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004013872310583938,
+      "loss": 2.9137,
+      "step": 12010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000401042248479057,
+      "loss": 2.8938,
+      "step": 12020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004006972658997201,
+      "loss": 2.9136,
+      "step": 12030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004003522833203833,
+      "loss": 2.9158,
+      "step": 12040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004000073007410465,
+      "loss": 2.9079,
+      "step": 12050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003996623181617096,
+      "loss": 2.9029,
+      "step": 12060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003993173355823728,
+      "loss": 2.9048,
+      "step": 12070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000398972353003036,
+      "loss": 2.9066,
+      "step": 12080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003986273704236992,
+      "loss": 2.8938,
+      "step": 12090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039828238784436233,
+      "loss": 2.9005,
+      "step": 12100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039793740526502554,
+      "loss": 2.8989,
+      "step": 12110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003975924226856887,
+      "loss": 2.8986,
+      "step": 12120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003972474401063519,
+      "loss": 2.9021,
+      "step": 12130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039690245752701506,
+      "loss": 2.898,
+      "step": 12140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039655747494767826,
+      "loss": 2.9101,
+      "step": 12150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003962124923683414,
+      "loss": 2.8961,
+      "step": 12160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003958675097890046,
+      "loss": 2.9234,
+      "step": 12170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003955225272096678,
+      "loss": 2.9072,
+      "step": 12180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000395177544630331,
+      "loss": 2.8984,
+      "step": 12190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039483256205099414,
+      "loss": 2.9119,
+      "step": 12200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039448757947165735,
+      "loss": 2.9038,
+      "step": 12210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003941425968923205,
+      "loss": 2.9036,
+      "step": 12220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039379761431298366,
+      "loss": 2.9056,
+      "step": 12230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039345263173364687,
+      "loss": 2.9175,
+      "step": 12240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039310764915431,
+      "loss": 2.8953,
+      "step": 12250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003927626665749732,
+      "loss": 2.9157,
+      "step": 12260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003924176839956364,
+      "loss": 2.9025,
+      "step": 12270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039207270141629954,
+      "loss": 2.9037,
+      "step": 12280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039172771883696274,
+      "loss": 2.9168,
+      "step": 12290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003913827362576259,
+      "loss": 2.9057,
+      "step": 12300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039103775367828905,
+      "loss": 2.9137,
+      "step": 12310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039069277109895226,
+      "loss": 2.9046,
+      "step": 12320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003903477885196154,
+      "loss": 2.8984,
+      "step": 12330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003900028059402786,
+      "loss": 2.9113,
+      "step": 12340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003896578233609418,
+      "loss": 2.8974,
+      "step": 12350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038931284078160493,
+      "loss": 2.9089,
+      "step": 12360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038896785820226814,
+      "loss": 2.8893,
+      "step": 12370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003886228756229313,
+      "loss": 2.8988,
+      "step": 12380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038827789304359445,
+      "loss": 2.903,
+      "step": 12390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00038793291046425766,
+      "loss": 2.902,
+      "step": 12400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003875879278849208,
+      "loss": 2.9099,
+      "step": 12410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000387242945305584,
+      "loss": 2.9157,
+      "step": 12420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003868979627262472,
+      "loss": 2.9171,
+      "step": 12430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003865529801469104,
+      "loss": 2.9036,
+      "step": 12440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003862079975675736,
+      "loss": 2.8988,
+      "step": 12450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038586301498823674,
+      "loss": 2.8979,
+      "step": 12460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038551803240889995,
+      "loss": 2.8959,
+      "step": 12470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003851730498295631,
+      "loss": 2.917,
+      "step": 12480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038482806725022626,
+      "loss": 2.9031,
+      "step": 12490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038448308467088947,
+      "loss": 2.8989,
+      "step": 12500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003841381020915526,
+      "loss": 2.8891,
+      "step": 12510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003837931195122158,
+      "loss": 2.8941,
+      "step": 12520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000383448136932879,
+      "loss": 2.9099,
+      "step": 12530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038310315435354214,
+      "loss": 2.8959,
+      "step": 12540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038275817177420534,
+      "loss": 2.9048,
+      "step": 12550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003824131891948685,
+      "loss": 2.8913,
+      "step": 12560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038206820661553165,
+      "loss": 2.8997,
+      "step": 12570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038172322403619486,
+      "loss": 2.9041,
+      "step": 12580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000381378241456858,
+      "loss": 2.8895,
+      "step": 12590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003810332588775212,
+      "loss": 2.9107,
+      "step": 12600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003806882762981844,
+      "loss": 2.8967,
+      "step": 12610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00038034329371884753,
+      "loss": 2.9056,
+      "step": 12620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037999831113951074,
+      "loss": 2.9141,
+      "step": 12630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003796533285601739,
+      "loss": 2.9218,
+      "step": 12640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037930834598083705,
+      "loss": 2.8984,
+      "step": 12650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037896336340150025,
+      "loss": 2.9125,
+      "step": 12660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003786183808221634,
+      "loss": 2.902,
+      "step": 12670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003782733982428266,
+      "loss": 2.9045,
+      "step": 12680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037792841566348977,
+      "loss": 2.8975,
+      "step": 12690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000377583433084153,
+      "loss": 2.8915,
+      "step": 12700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003772384505048162,
+      "loss": 2.9082,
+      "step": 12710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037689346792547934,
+      "loss": 2.9201,
+      "step": 12720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037654848534614255,
+      "loss": 2.8956,
+      "step": 12730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003762035027668057,
+      "loss": 2.915,
+      "step": 12740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037585852018746886,
+      "loss": 2.9026,
+      "step": 12750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037551353760813206,
+      "loss": 2.9105,
+      "step": 12760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003751685550287952,
+      "loss": 2.8919,
+      "step": 12770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037482357244945837,
+      "loss": 2.9143,
+      "step": 12780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003744785898701216,
+      "loss": 2.9003,
+      "step": 12790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037413360729078473,
+      "loss": 2.912,
+      "step": 12800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037378862471144794,
+      "loss": 2.8943,
+      "step": 12810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003734436421321111,
+      "loss": 2.8961,
+      "step": 12820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037309865955277425,
+      "loss": 2.9033,
+      "step": 12830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037275367697343746,
+      "loss": 2.9034,
+      "step": 12840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003724086943941006,
+      "loss": 2.9188,
+      "step": 12850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003720637118147638,
+      "loss": 2.8961,
+      "step": 12860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000371718729235427,
+      "loss": 2.9117,
+      "step": 12870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037137374665609013,
+      "loss": 2.912,
+      "step": 12880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037102876407675334,
+      "loss": 2.8985,
+      "step": 12890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003706837814974165,
+      "loss": 2.9125,
+      "step": 12900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003703387989180797,
+      "loss": 2.9077,
+      "step": 12910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036999381633874285,
+      "loss": 2.898,
+      "step": 12920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000369648833759406,
+      "loss": 2.8929,
+      "step": 12930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003693038511800692,
+      "loss": 2.9012,
+      "step": 12940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036895886860073237,
+      "loss": 2.9158,
+      "step": 12950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003686138860213956,
+      "loss": 2.9175,
+      "step": 12960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036826890344205873,
+      "loss": 2.9128,
+      "step": 12970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036792392086272194,
+      "loss": 2.9037,
+      "step": 12980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036757893828338515,
+      "loss": 2.906,
+      "step": 12990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003672339557040483,
+      "loss": 2.9004,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036688897312471146,
+      "loss": 2.9017,
+      "step": 13010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036654399054537466,
+      "loss": 2.8926,
+      "step": 13020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003661990079660378,
+      "loss": 2.8996,
+      "step": 13030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000365854025386701,
+      "loss": 2.909,
+      "step": 13040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003655090428073642,
+      "loss": 2.8995,
+      "step": 13050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036516406022802733,
+      "loss": 2.9105,
+      "step": 13060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036481907764869054,
+      "loss": 2.9052,
+      "step": 13070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003644740950693537,
+      "loss": 2.896,
+      "step": 13080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036412911249001685,
+      "loss": 2.9137,
+      "step": 13090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00036378412991068006,
+      "loss": 2.8978,
+      "step": 13100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003634391473313432,
+      "loss": 2.8995,
+      "step": 13110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003630941647520064,
+      "loss": 2.9088,
+      "step": 13120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003627491821726696,
+      "loss": 2.912,
+      "step": 13130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00036240419959333273,
+      "loss": 2.8979,
+      "step": 13140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00036205921701399594,
+      "loss": 2.8966,
+      "step": 13150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003617142344346591,
+      "loss": 2.9058,
+      "step": 13160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003613692518553223,
+      "loss": 2.9026,
+      "step": 13170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00036102426927598545,
+      "loss": 2.903,
+      "step": 13180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003606792866966486,
+      "loss": 2.9088,
+      "step": 13190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003603343041173118,
+      "loss": 2.9121,
+      "step": 13200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035998932153797497,
+      "loss": 2.8987,
+      "step": 13210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003596443389586382,
+      "loss": 2.9055,
+      "step": 13220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035929935637930133,
+      "loss": 2.8996,
+      "step": 13230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035895437379996454,
+      "loss": 2.9042,
+      "step": 13240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003586093912206277,
+      "loss": 2.8981,
+      "step": 13250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003582644086412909,
+      "loss": 2.9012,
+      "step": 13260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035791942606195405,
+      "loss": 2.9027,
+      "step": 13270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035757444348261726,
+      "loss": 2.8897,
+      "step": 13280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003572294609032804,
+      "loss": 2.8932,
+      "step": 13290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003568844783239436,
+      "loss": 2.91,
+      "step": 13300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003565394957446068,
+      "loss": 2.8957,
+      "step": 13310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035619451316526993,
+      "loss": 2.9059,
+      "step": 13320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035584953058593314,
+      "loss": 2.8968,
+      "step": 13330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003555045480065963,
+      "loss": 2.9006,
+      "step": 13340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00035515956542725945,
+      "loss": 2.8971,
+      "step": 13350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00035481458284792266,
+      "loss": 2.8986,
+      "step": 13360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003544696002685858,
+      "loss": 2.9123,
+      "step": 13370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000354124617689249,
+      "loss": 2.9113,
+      "step": 13380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003537796351099122,
+      "loss": 2.9019,
+      "step": 13390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00035343465253057533,
+      "loss": 2.8994,
+      "step": 13400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00035308966995123854,
+      "loss": 2.8933,
+      "step": 13410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003527446873719017,
+      "loss": 2.8987,
+      "step": 13420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003523997047925649,
+      "loss": 2.9023,
+      "step": 13430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00035205472221322805,
+      "loss": 2.9153,
+      "step": 13440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003517097396338912,
+      "loss": 2.904,
+      "step": 13450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003513647570545544,
+      "loss": 2.8995,
+      "step": 13460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00035101977447521757,
+      "loss": 2.9052,
+      "step": 13470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003506747918958807,
+      "loss": 2.9099,
+      "step": 13480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00035032980931654393,
+      "loss": 2.8962,
+      "step": 13490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034998482673720714,
+      "loss": 2.9076,
+      "step": 13500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003496398441578703,
+      "loss": 2.9014,
+      "step": 13510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003492948615785335,
+      "loss": 2.8989,
+      "step": 13520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034894987899919665,
+      "loss": 2.9014,
+      "step": 13530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034860489641985986,
+      "loss": 2.9207,
+      "step": 13540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000348259913840523,
+      "loss": 2.9052,
+      "step": 13550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003479149312611862,
+      "loss": 2.9101,
+      "step": 13560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003475699486818494,
+      "loss": 2.9098,
+      "step": 13570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034722496610251253,
+      "loss": 2.9102,
+      "step": 13580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00034687998352317574,
+      "loss": 2.9082,
+      "step": 13590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003465350009438389,
+      "loss": 2.9105,
+      "step": 13600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034619001836450205,
+      "loss": 2.8933,
+      "step": 13610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034584503578516526,
+      "loss": 2.9006,
+      "step": 13620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003455000532058284,
+      "loss": 2.8992,
+      "step": 13630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003451550706264916,
+      "loss": 2.9023,
+      "step": 13640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034481008804715477,
+      "loss": 2.888,
+      "step": 13650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003444651054678179,
+      "loss": 2.9107,
+      "step": 13660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034412012288848113,
+      "loss": 2.8991,
+      "step": 13670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003437751403091443,
+      "loss": 2.8979,
+      "step": 13680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003434301577298075,
+      "loss": 2.9068,
+      "step": 13690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034308517515047065,
+      "loss": 2.8981,
+      "step": 13700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003427401925711338,
+      "loss": 2.9153,
+      "step": 13710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000342395209991797,
+      "loss": 2.8937,
+      "step": 13720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034205022741246017,
+      "loss": 2.8903,
+      "step": 13730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003417052448331233,
+      "loss": 2.9081,
+      "step": 13740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00034136026225378653,
+      "loss": 2.8954,
+      "step": 13750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003410152796744497,
+      "loss": 2.8996,
+      "step": 13760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003406702970951129,
+      "loss": 2.9184,
+      "step": 13770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003403253145157761,
+      "loss": 2.9027,
+      "step": 13780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033998033193643925,
+      "loss": 2.907,
+      "step": 13790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00033963534935710246,
+      "loss": 2.9031,
+      "step": 13800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003392903667777656,
+      "loss": 2.898,
+      "step": 13810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003389453841984288,
+      "loss": 2.8932,
+      "step": 13820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000338600401619092,
+      "loss": 2.8943,
+      "step": 13830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033825541903975513,
+      "loss": 2.9122,
+      "step": 13840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033791043646041834,
+      "loss": 2.8998,
+      "step": 13850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003375654538810815,
+      "loss": 2.9157,
+      "step": 13860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033722047130174465,
+      "loss": 2.9032,
+      "step": 13870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033687548872240786,
+      "loss": 2.9028,
+      "step": 13880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000336530506143071,
+      "loss": 2.906,
+      "step": 13890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003361855235637342,
+      "loss": 2.9011,
+      "step": 13900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033584054098439737,
+      "loss": 2.903,
+      "step": 13910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003354955584050605,
+      "loss": 2.9047,
+      "step": 13920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033515057582572373,
+      "loss": 2.9057,
+      "step": 13930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003348055932463869,
+      "loss": 2.9092,
+      "step": 13940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003344606106670501,
+      "loss": 2.9024,
+      "step": 13950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033411562808771325,
+      "loss": 2.9131,
+      "step": 13960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003337706455083764,
+      "loss": 2.8908,
+      "step": 13970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003334256629290396,
+      "loss": 2.8965,
+      "step": 13980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033308068034970277,
+      "loss": 2.9001,
+      "step": 13990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003327356977703659,
+      "loss": 2.8903,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033239071519102913,
+      "loss": 2.9075,
+      "step": 14010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003320457326116923,
+      "loss": 2.9053,
+      "step": 14020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003317007500323555,
+      "loss": 2.9011,
+      "step": 14030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033135576745301864,
+      "loss": 2.9048,
+      "step": 14040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033101078487368185,
+      "loss": 2.9141,
+      "step": 14050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00033066580229434506,
+      "loss": 2.9109,
+      "step": 14060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003303208197150082,
+      "loss": 2.9088,
+      "step": 14070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003299758371356714,
+      "loss": 2.9116,
+      "step": 14080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003296308545563346,
+      "loss": 2.9162,
+      "step": 14090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032928587197699773,
+      "loss": 2.9033,
+      "step": 14100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032894088939766094,
+      "loss": 2.9128,
+      "step": 14110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003285959068183241,
+      "loss": 2.905,
+      "step": 14120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032825092423898725,
+      "loss": 2.9114,
+      "step": 14130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032790594165965045,
+      "loss": 2.9041,
+      "step": 14140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003275609590803136,
+      "loss": 2.9176,
+      "step": 14150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003272159765009768,
+      "loss": 2.9039,
+      "step": 14160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032687099392163997,
+      "loss": 2.8964,
+      "step": 14170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003265260113423031,
+      "loss": 2.8992,
+      "step": 14180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032618102876296633,
+      "loss": 2.905,
+      "step": 14190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003258360461836295,
+      "loss": 2.9047,
+      "step": 14200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003254910636042927,
+      "loss": 2.9108,
+      "step": 14210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032514608102495585,
+      "loss": 2.8989,
+      "step": 14220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000324801098445619,
+      "loss": 2.9076,
+      "step": 14230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003244561158662822,
+      "loss": 2.9156,
+      "step": 14240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032411113328694537,
+      "loss": 2.9069,
+      "step": 14250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003237661507076085,
+      "loss": 2.9098,
+      "step": 14260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032342116812827173,
+      "loss": 2.904,
+      "step": 14270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003230761855489349,
+      "loss": 2.8941,
+      "step": 14280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003227312029695981,
+      "loss": 2.89,
+      "step": 14290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00032238622039026124,
+      "loss": 2.9128,
+      "step": 14300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00032204123781092445,
+      "loss": 2.894,
+      "step": 14310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003216962552315876,
+      "loss": 2.8963,
+      "step": 14320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003213512726522508,
+      "loss": 2.9093,
+      "step": 14330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00032100629007291397,
+      "loss": 2.9001,
+      "step": 14340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003206613074935772,
+      "loss": 2.8994,
+      "step": 14350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00032031632491424033,
+      "loss": 2.899,
+      "step": 14360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031997134233490354,
+      "loss": 2.8903,
+      "step": 14370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003196263597555667,
+      "loss": 2.9087,
+      "step": 14380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031928137717622985,
+      "loss": 2.9041,
+      "step": 14390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031893639459689305,
+      "loss": 2.9007,
+      "step": 14400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003185914120175562,
+      "loss": 2.909,
+      "step": 14410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003182464294382194,
+      "loss": 2.914,
+      "step": 14420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031790144685888257,
+      "loss": 2.9128,
+      "step": 14430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003175564642795457,
+      "loss": 2.8977,
+      "step": 14440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031721148170020893,
+      "loss": 2.9093,
+      "step": 14450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003168664991208721,
+      "loss": 2.8941,
+      "step": 14460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003165215165415353,
+      "loss": 2.9142,
+      "step": 14470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031617653396219845,
+      "loss": 2.9167,
+      "step": 14480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003158315513828616,
+      "loss": 2.8878,
+      "step": 14490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003154865688035248,
+      "loss": 2.9121,
+      "step": 14500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00031514158622418796,
+      "loss": 2.8967,
+      "step": 14510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003147966036448511,
+      "loss": 2.8937,
+      "step": 14520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003144516210655143,
+      "loss": 2.9008,
+      "step": 14530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003141066384861775,
+      "loss": 2.9129,
+      "step": 14540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003137616559068407,
+      "loss": 2.9162,
+      "step": 14550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00031341667332750384,
+      "loss": 2.902,
+      "step": 14560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00031307169074816705,
+      "loss": 2.894,
+      "step": 14570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003127267081688302,
+      "loss": 2.9008,
+      "step": 14580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003123817255894934,
+      "loss": 2.9101,
+      "step": 14590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00031203674301015657,
+      "loss": 2.906,
+      "step": 14600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003116917604308198,
+      "loss": 2.9022,
+      "step": 14610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00031134677785148293,
+      "loss": 2.893,
+      "step": 14620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00031100179527214614,
+      "loss": 2.8987,
+      "step": 14630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003106568126928093,
+      "loss": 2.9177,
+      "step": 14640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00031031183011347244,
+      "loss": 2.9007,
+      "step": 14650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030996684753413565,
+      "loss": 2.8981,
+      "step": 14660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003096218649547988,
+      "loss": 2.899,
+      "step": 14670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000309276882375462,
+      "loss": 2.9034,
+      "step": 14680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030893189979612517,
+      "loss": 2.9074,
+      "step": 14690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003085869172167883,
+      "loss": 2.9148,
+      "step": 14700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030824193463745153,
+      "loss": 2.8953,
+      "step": 14710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003078969520581147,
+      "loss": 2.8986,
+      "step": 14720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003075519694787779,
+      "loss": 2.9074,
+      "step": 14730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030720698689944105,
+      "loss": 2.9074,
+      "step": 14740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003068620043201042,
+      "loss": 2.9034,
+      "step": 14750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003065170217407674,
+      "loss": 2.9011,
+      "step": 14760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00030617203916143056,
+      "loss": 2.9044,
+      "step": 14770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003058270565820937,
+      "loss": 2.9048,
+      "step": 14780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003054820740027569,
+      "loss": 2.9206,
+      "step": 14790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003051370914234201,
+      "loss": 2.8992,
+      "step": 14800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003047921088440833,
+      "loss": 2.8983,
+      "step": 14810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030444712626474644,
+      "loss": 2.8909,
+      "step": 14820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003041021436854096,
+      "loss": 2.9059,
+      "step": 14830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003037571611060728,
+      "loss": 2.9049,
+      "step": 14840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030341217852673596,
+      "loss": 2.893,
+      "step": 14850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030306719594739917,
+      "loss": 2.9065,
+      "step": 14860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003027222133680624,
+      "loss": 2.9091,
+      "step": 14870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030237723078872553,
+      "loss": 2.8959,
+      "step": 14880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030203224820938874,
+      "loss": 2.8941,
+      "step": 14890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003016872656300519,
+      "loss": 2.9125,
+      "step": 14900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030134228305071504,
+      "loss": 2.8967,
+      "step": 14910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00030099730047137825,
+      "loss": 2.8891,
+      "step": 14920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003006523178920414,
+      "loss": 2.9021,
+      "step": 14930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003003073353127046,
+      "loss": 2.9078,
+      "step": 14940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029996235273336777,
+      "loss": 2.894,
+      "step": 14950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002996173701540309,
+      "loss": 2.9118,
+      "step": 14960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029927238757469413,
+      "loss": 2.9143,
+      "step": 14970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002989274049953573,
+      "loss": 2.8984,
+      "step": 14980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002985824224160205,
+      "loss": 2.9027,
+      "step": 14990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029823743983668365,
+      "loss": 2.9083,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002978924572573468,
+      "loss": 2.9187,
+      "step": 15010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029754747467801,
+      "loss": 2.9013,
+      "step": 15020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029720249209867316,
+      "loss": 2.9094,
+      "step": 15030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002968575095193363,
+      "loss": 2.8991,
+      "step": 15040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002965125269399995,
+      "loss": 2.9133,
+      "step": 15050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002961675443606627,
+      "loss": 2.9101,
+      "step": 15060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002958225617813259,
+      "loss": 2.9082,
+      "step": 15070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029547757920198904,
+      "loss": 2.9103,
+      "step": 15080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002951325966226522,
+      "loss": 2.9063,
+      "step": 15090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002947876140433154,
+      "loss": 2.8965,
+      "step": 15100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029444263146397856,
+      "loss": 2.9041,
+      "step": 15110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029409764888464176,
+      "loss": 2.9085,
+      "step": 15120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002937526663053049,
+      "loss": 2.9023,
+      "step": 15130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002934076837259681,
+      "loss": 2.9071,
+      "step": 15140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029306270114663134,
+      "loss": 2.9177,
+      "step": 15150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002927177185672945,
+      "loss": 2.892,
+      "step": 15160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029237273598795764,
+      "loss": 2.9077,
+      "step": 15170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029202775340862085,
+      "loss": 2.9043,
+      "step": 15180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000291682770829284,
+      "loss": 2.9068,
+      "step": 15190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002913377882499472,
+      "loss": 2.9058,
+      "step": 15200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029099280567061037,
+      "loss": 2.9103,
+      "step": 15210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002906478230912735,
+      "loss": 2.8996,
+      "step": 15220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029030284051193673,
+      "loss": 2.9062,
+      "step": 15230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002899578579325999,
+      "loss": 2.9052,
+      "step": 15240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896128753532631,
+      "loss": 2.9025,
+      "step": 15250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028926789277392625,
+      "loss": 2.8958,
+      "step": 15260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002889229101945894,
+      "loss": 2.9085,
+      "step": 15270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002885779276152526,
+      "loss": 2.895,
+      "step": 15280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028823294503591576,
+      "loss": 2.8991,
+      "step": 15290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002878879624565789,
+      "loss": 2.8967,
+      "step": 15300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002875429798772421,
+      "loss": 2.9001,
+      "step": 15310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002871979972979053,
+      "loss": 2.9197,
+      "step": 15320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002868530147185685,
+      "loss": 2.902,
+      "step": 15330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028650803213923164,
+      "loss": 2.917,
+      "step": 15340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002861630495598948,
+      "loss": 2.9009,
+      "step": 15350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000285818066980558,
+      "loss": 2.8979,
+      "step": 15360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028547308440122116,
+      "loss": 2.8964,
+      "step": 15370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028512810182188436,
+      "loss": 2.9035,
+      "step": 15380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002847831192425475,
+      "loss": 2.9096,
+      "step": 15390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002844381366632107,
+      "loss": 2.8931,
+      "step": 15400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002840931540838739,
+      "loss": 2.917,
+      "step": 15410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002837481715045371,
+      "loss": 2.9025,
+      "step": 15420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028340318892520024,
+      "loss": 2.9047,
+      "step": 15430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028305820634586345,
+      "loss": 2.9177,
+      "step": 15440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002827132237665266,
+      "loss": 2.9076,
+      "step": 15450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002823682411871898,
+      "loss": 2.904,
+      "step": 15460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028202325860785297,
+      "loss": 2.8995,
+      "step": 15470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002816782760285161,
+      "loss": 2.8939,
+      "step": 15480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028133329344917933,
+      "loss": 2.891,
+      "step": 15490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002809883108698425,
+      "loss": 2.9038,
+      "step": 15500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002806433282905057,
+      "loss": 2.8983,
+      "step": 15510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028029834571116884,
+      "loss": 2.9036,
+      "step": 15520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000279953363131832,
+      "loss": 2.8955,
+      "step": 15530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002796083805524952,
+      "loss": 2.9102,
+      "step": 15540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027926339797315836,
+      "loss": 2.8984,
+      "step": 15550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002789184153938215,
+      "loss": 2.8999,
+      "step": 15560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002785734328144847,
+      "loss": 2.907,
+      "step": 15570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002782284502351479,
+      "loss": 2.8988,
+      "step": 15580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002778834676558111,
+      "loss": 2.9042,
+      "step": 15590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027753848507647424,
+      "loss": 2.8921,
+      "step": 15600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002771935024971374,
+      "loss": 2.9041,
+      "step": 15610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002768485199178006,
+      "loss": 2.8948,
+      "step": 15620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027650353733846376,
+      "loss": 2.9202,
+      "step": 15630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027615855475912696,
+      "loss": 2.8977,
+      "step": 15640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002758135721797901,
+      "loss": 2.9125,
+      "step": 15650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002754685896004533,
+      "loss": 2.9018,
+      "step": 15660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002751236070211165,
+      "loss": 2.9056,
+      "step": 15670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002747786244417797,
+      "loss": 2.899,
+      "step": 15680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027443364186244284,
+      "loss": 2.9026,
+      "step": 15690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027408865928310605,
+      "loss": 2.9064,
+      "step": 15700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002737436767037692,
+      "loss": 2.9086,
+      "step": 15710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002733986941244324,
+      "loss": 2.9036,
+      "step": 15720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00027305371154509557,
+      "loss": 2.8941,
+      "step": 15730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002727087289657587,
+      "loss": 2.9115,
+      "step": 15740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00027236374638642193,
+      "loss": 2.9032,
+      "step": 15750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002720187638070851,
+      "loss": 2.9025,
+      "step": 15760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002716737812277483,
+      "loss": 2.8988,
+      "step": 15770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00027132879864841144,
+      "loss": 2.9002,
+      "step": 15780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002709838160690746,
+      "loss": 2.8966,
+      "step": 15790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002706388334897378,
+      "loss": 2.9004,
+      "step": 15800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00027029385091040096,
+      "loss": 2.8962,
+      "step": 15810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002699488683310641,
+      "loss": 2.9054,
+      "step": 15820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002696038857517273,
+      "loss": 2.908,
+      "step": 15830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002692589031723905,
+      "loss": 2.9048,
+      "step": 15840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002689139205930537,
+      "loss": 2.9087,
+      "step": 15850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026856893801371684,
+      "loss": 2.8946,
+      "step": 15860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026822395543438,
+      "loss": 2.9033,
+      "step": 15870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002678789728550432,
+      "loss": 2.9062,
+      "step": 15880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026753399027570635,
+      "loss": 2.9029,
+      "step": 15890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026718900769636956,
+      "loss": 2.9044,
+      "step": 15900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002668440251170327,
+      "loss": 2.9028,
+      "step": 15910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026649904253769587,
+      "loss": 2.9098,
+      "step": 15920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002661540599583591,
+      "loss": 2.9037,
+      "step": 15930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002658090773790223,
+      "loss": 2.9033,
+      "step": 15940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026546409479968544,
+      "loss": 2.8988,
+      "step": 15950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026511911222034865,
+      "loss": 2.9087,
+      "step": 15960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002647741296410118,
+      "loss": 2.8993,
+      "step": 15970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000264429147061675,
+      "loss": 2.9117,
+      "step": 15980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026408416448233816,
+      "loss": 2.9083,
+      "step": 15990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002637391819030013,
+      "loss": 2.922,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002633941993236645,
+      "loss": 2.9061,
+      "step": 16010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002630492167443277,
+      "loss": 2.9005,
+      "step": 16020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002627042341649909,
+      "loss": 2.9011,
+      "step": 16030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026235925158565404,
+      "loss": 2.8957,
+      "step": 16040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002620142690063172,
+      "loss": 2.9079,
+      "step": 16050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002616692864269804,
+      "loss": 2.8908,
+      "step": 16060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026132430384764356,
+      "loss": 2.9011,
+      "step": 16070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002609793212683067,
+      "loss": 2.8975,
+      "step": 16080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002606343386889699,
+      "loss": 2.8969,
+      "step": 16090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002602893561096331,
+      "loss": 2.908,
+      "step": 16100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002599443735302963,
+      "loss": 2.9007,
+      "step": 16110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025959939095095944,
+      "loss": 2.896,
+      "step": 16120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002592544083716226,
+      "loss": 2.9162,
+      "step": 16130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002589094257922858,
+      "loss": 2.9127,
+      "step": 16140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025856444321294895,
+      "loss": 2.8977,
+      "step": 16150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025821946063361216,
+      "loss": 2.8996,
+      "step": 16160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002578744780542753,
+      "loss": 2.9122,
+      "step": 16170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025752949547493847,
+      "loss": 2.9109,
+      "step": 16180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002571845128956017,
+      "loss": 2.8954,
+      "step": 16190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025683953031626483,
+      "loss": 2.8955,
+      "step": 16200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025649454773692804,
+      "loss": 2.9147,
+      "step": 16210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025614956515759125,
+      "loss": 2.9029,
+      "step": 16220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002558045825782544,
+      "loss": 2.9036,
+      "step": 16230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002554595999989176,
+      "loss": 2.9001,
+      "step": 16240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025511461741958076,
+      "loss": 2.9073,
+      "step": 16250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002547696348402439,
+      "loss": 2.9172,
+      "step": 16260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002544246522609071,
+      "loss": 2.9078,
+      "step": 16270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002540796696815703,
+      "loss": 2.908,
+      "step": 16280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002537346871022335,
+      "loss": 2.8922,
+      "step": 16290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025338970452289664,
+      "loss": 2.897,
+      "step": 16300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002530447219435598,
+      "loss": 2.906,
+      "step": 16310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000252699739364223,
+      "loss": 2.9161,
+      "step": 16320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025235475678488616,
+      "loss": 2.9104,
+      "step": 16330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002520097742055493,
+      "loss": 2.9036,
+      "step": 16340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002516647916262125,
+      "loss": 2.9064,
+      "step": 16350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002513198090468757,
+      "loss": 2.9027,
+      "step": 16360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002509748264675389,
+      "loss": 2.9073,
+      "step": 16370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025062984388820204,
+      "loss": 2.8976,
+      "step": 16380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002502848613088652,
+      "loss": 2.9061,
+      "step": 16390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002499398787295284,
+      "loss": 2.901,
+      "step": 16400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024959489615019155,
+      "loss": 2.9059,
+      "step": 16410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00024924991357085476,
+      "loss": 2.8972,
+      "step": 16420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002489049309915179,
+      "loss": 2.9037,
+      "step": 16430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024855994841218107,
+      "loss": 2.9032,
+      "step": 16440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002482149658328443,
+      "loss": 2.9265,
+      "step": 16450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024786998325350743,
+      "loss": 2.9022,
+      "step": 16460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024752500067417064,
+      "loss": 2.9097,
+      "step": 16470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002471800180948338,
+      "loss": 2.9039,
+      "step": 16480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000246835035515497,
+      "loss": 2.9017,
+      "step": 16490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002464900529361602,
+      "loss": 2.9121,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024614507035682336,
+      "loss": 2.9075,
+      "step": 16510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002458000877774865,
+      "loss": 2.9112,
+      "step": 16520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002454551051981497,
+      "loss": 2.9222,
+      "step": 16530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002451101226188129,
+      "loss": 2.9017,
+      "step": 16540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002447651400394761,
+      "loss": 2.8949,
+      "step": 16550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024442015746013924,
+      "loss": 2.8825,
+      "step": 16560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002440751748808024,
+      "loss": 2.8994,
+      "step": 16570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002437301923014656,
+      "loss": 2.9226,
+      "step": 16580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024338520972212876,
+      "loss": 2.8979,
+      "step": 16590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002430402271427919,
+      "loss": 2.9146,
+      "step": 16600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024269524456345512,
+      "loss": 2.9017,
+      "step": 16610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024235026198411827,
+      "loss": 2.9028,
+      "step": 16620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024200527940478148,
+      "loss": 2.9051,
+      "step": 16630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00024166029682544464,
+      "loss": 2.9023,
+      "step": 16640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002413153142461078,
+      "loss": 2.9008,
+      "step": 16650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000240970331666771,
+      "loss": 2.9097,
+      "step": 16660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00024062534908743415,
+      "loss": 2.889,
+      "step": 16670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00024028036650809736,
+      "loss": 2.9022,
+      "step": 16680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023993538392876054,
+      "loss": 2.9029,
+      "step": 16690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002395904013494237,
+      "loss": 2.9054,
+      "step": 16700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002392454187700869,
+      "loss": 2.903,
+      "step": 16710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023890043619075006,
+      "loss": 2.9106,
+      "step": 16720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002385554536114132,
+      "loss": 2.904,
+      "step": 16730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023821047103207642,
+      "loss": 2.9038,
+      "step": 16740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023786548845273957,
+      "loss": 2.9098,
+      "step": 16750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023752050587340278,
+      "loss": 2.9109,
+      "step": 16760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023717552329406594,
+      "loss": 2.9062,
+      "step": 16770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002368305407147291,
+      "loss": 2.9098,
+      "step": 16780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002364855581353923,
+      "loss": 2.9194,
+      "step": 16790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023614057555605545,
+      "loss": 2.9046,
+      "step": 16800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023579559297671866,
+      "loss": 2.9114,
+      "step": 16810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002354506103973818,
+      "loss": 2.9039,
+      "step": 16820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000235105627818045,
+      "loss": 2.9002,
+      "step": 16830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002347606452387082,
+      "loss": 2.9103,
+      "step": 16840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023441566265937136,
+      "loss": 2.9085,
+      "step": 16850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002340706800800345,
+      "loss": 2.9041,
+      "step": 16860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023372569750069772,
+      "loss": 2.9011,
+      "step": 16870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023338071492136087,
+      "loss": 2.9164,
+      "step": 16880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023303573234202408,
+      "loss": 2.9072,
+      "step": 16890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023269074976268723,
+      "loss": 2.8972,
+      "step": 16900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002323457671833504,
+      "loss": 2.8869,
+      "step": 16910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002320007846040136,
+      "loss": 2.9,
+      "step": 16920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00023165580202467675,
+      "loss": 2.9002,
+      "step": 16930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00023131081944533996,
+      "loss": 2.9002,
+      "step": 16940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002309658368660031,
+      "loss": 2.8862,
+      "step": 16950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002306208542866663,
+      "loss": 2.9052,
+      "step": 16960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002302758717073295,
+      "loss": 2.8878,
+      "step": 16970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022993088912799266,
+      "loss": 2.9049,
+      "step": 16980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002295859065486558,
+      "loss": 2.8984,
+      "step": 16990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022924092396931902,
+      "loss": 2.9106,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022889594138998217,
+      "loss": 2.9045,
+      "step": 17010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022855095881064538,
+      "loss": 2.8971,
+      "step": 17020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022820597623130853,
+      "loss": 2.8969,
+      "step": 17030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002278609936519717,
+      "loss": 2.9046,
+      "step": 17040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002275160110726349,
+      "loss": 2.9112,
+      "step": 17050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022717102849329805,
+      "loss": 2.9069,
+      "step": 17060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022682604591396126,
+      "loss": 2.9111,
+      "step": 17070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002264810633346244,
+      "loss": 2.9147,
+      "step": 17080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002261360807552876,
+      "loss": 2.9128,
+      "step": 17090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022579109817595077,
+      "loss": 2.904,
+      "step": 17100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022544611559661396,
+      "loss": 2.9082,
+      "step": 17110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002251011330172771,
+      "loss": 2.908,
+      "step": 17120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022475615043794032,
+      "loss": 2.9086,
+      "step": 17130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022441116785860347,
+      "loss": 2.8999,
+      "step": 17140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022406618527926668,
+      "loss": 2.9021,
+      "step": 17150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022372120269992983,
+      "loss": 2.8954,
+      "step": 17160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000223376220120593,
+      "loss": 2.9067,
+      "step": 17170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002230312375412562,
+      "loss": 2.8907,
+      "step": 17180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022268625496191935,
+      "loss": 2.9144,
+      "step": 17190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022234127238258256,
+      "loss": 2.8979,
+      "step": 17200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002219962898032457,
+      "loss": 2.9123,
+      "step": 17210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002216513072239089,
+      "loss": 2.9021,
+      "step": 17220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022130632464457207,
+      "loss": 2.9224,
+      "step": 17230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022096134206523526,
+      "loss": 2.9068,
+      "step": 17240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002206163594858984,
+      "loss": 2.9137,
+      "step": 17250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022027137690656162,
+      "loss": 2.8901,
+      "step": 17260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021992639432722477,
+      "loss": 2.9055,
+      "step": 17270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021958141174788798,
+      "loss": 2.8794,
+      "step": 17280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021923642916855113,
+      "loss": 2.9054,
+      "step": 17290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002188914465892143,
+      "loss": 2.8954,
+      "step": 17300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002185464640098775,
+      "loss": 2.9107,
+      "step": 17310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021820148143054065,
+      "loss": 2.8976,
+      "step": 17320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021785649885120386,
+      "loss": 2.8968,
+      "step": 17330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000217511516271867,
+      "loss": 2.9104,
+      "step": 17340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002171665336925302,
+      "loss": 2.9066,
+      "step": 17350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021682155111319337,
+      "loss": 2.8949,
+      "step": 17360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00021647656853385655,
+      "loss": 2.9018,
+      "step": 17370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002161315859545197,
+      "loss": 2.8992,
+      "step": 17380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021578660337518292,
+      "loss": 2.9018,
+      "step": 17390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021544162079584607,
+      "loss": 2.9019,
+      "step": 17400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021509663821650928,
+      "loss": 2.9091,
+      "step": 17410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021475165563717243,
+      "loss": 2.9233,
+      "step": 17420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002144066730578356,
+      "loss": 2.9097,
+      "step": 17430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002140616904784988,
+      "loss": 2.8883,
+      "step": 17440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021371670789916195,
+      "loss": 2.9116,
+      "step": 17450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021337172531982516,
+      "loss": 2.906,
+      "step": 17460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002130267427404883,
+      "loss": 2.9042,
+      "step": 17470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002126817601611515,
+      "loss": 2.9129,
+      "step": 17480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021233677758181467,
+      "loss": 2.8952,
+      "step": 17490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021199179500247785,
+      "loss": 2.9107,
+      "step": 17500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000211646812423141,
+      "loss": 2.9124,
+      "step": 17510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021130182984380422,
+      "loss": 2.905,
+      "step": 17520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021095684726446737,
+      "loss": 2.9213,
+      "step": 17530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021061186468513058,
+      "loss": 2.9049,
+      "step": 17540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00021026688210579373,
+      "loss": 2.9091,
+      "step": 17550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002099218995264569,
+      "loss": 2.9056,
+      "step": 17560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002095769169471201,
+      "loss": 2.9074,
+      "step": 17570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020923193436778325,
+      "loss": 2.9103,
+      "step": 17580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020888695178844646,
+      "loss": 2.9109,
+      "step": 17590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002085419692091096,
+      "loss": 2.8959,
+      "step": 17600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00020819698662977276,
+      "loss": 2.8879,
+      "step": 17610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020785200405043597,
+      "loss": 2.9199,
+      "step": 17620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020750702147109915,
+      "loss": 2.9023,
+      "step": 17630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002071620388917623,
+      "loss": 2.9057,
+      "step": 17640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020681705631242552,
+      "loss": 2.9028,
+      "step": 17650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020647207373308867,
+      "loss": 2.9062,
+      "step": 17660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020612709115375188,
+      "loss": 2.9033,
+      "step": 17670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020578210857441503,
+      "loss": 2.9109,
+      "step": 17680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020543712599507819,
+      "loss": 2.9013,
+      "step": 17690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002050921434157414,
+      "loss": 2.8958,
+      "step": 17700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020474716083640455,
+      "loss": 2.8996,
+      "step": 17710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020440217825706776,
+      "loss": 2.9098,
+      "step": 17720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002040571956777309,
+      "loss": 2.8966,
+      "step": 17730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020371221309839406,
+      "loss": 2.9085,
+      "step": 17740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020336723051905727,
+      "loss": 2.8904,
+      "step": 17750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020302224793972045,
+      "loss": 2.913,
+      "step": 17760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002026772653603836,
+      "loss": 2.9138,
+      "step": 17770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020233228278104682,
+      "loss": 2.8912,
+      "step": 17780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020198730020170997,
+      "loss": 2.8997,
+      "step": 17790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020164231762237318,
+      "loss": 2.9092,
+      "step": 17800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020129733504303633,
+      "loss": 2.8924,
+      "step": 17810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020095235246369949,
+      "loss": 2.8965,
+      "step": 17820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002006073698843627,
+      "loss": 2.9135,
+      "step": 17830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00020026238730502585,
+      "loss": 2.9047,
+      "step": 17840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019991740472568906,
+      "loss": 2.8978,
+      "step": 17850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001995724221463522,
+      "loss": 2.9071,
+      "step": 17860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001992274395670154,
+      "loss": 2.9052,
+      "step": 17870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019888245698767854,
+      "loss": 2.9082,
+      "step": 17880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019853747440834173,
+      "loss": 2.9189,
+      "step": 17890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001981924918290049,
+      "loss": 2.9064,
+      "step": 17900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019784750924966812,
+      "loss": 2.8918,
+      "step": 17910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019750252667033127,
+      "loss": 2.9054,
+      "step": 17920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019715754409099445,
+      "loss": 2.8852,
+      "step": 17930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019681256151165763,
+      "loss": 2.9018,
+      "step": 17940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001964675789323208,
+      "loss": 2.9076,
+      "step": 17950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000196122596352984,
+      "loss": 2.9095,
+      "step": 17960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019577761377364715,
+      "loss": 2.9076,
+      "step": 17970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019543263119431033,
+      "loss": 2.9096,
+      "step": 17980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001950876486149735,
+      "loss": 2.9087,
+      "step": 17990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001947426660356367,
+      "loss": 2.9141,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019439768345629984,
+      "loss": 2.9057,
+      "step": 18010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019405270087696303,
+      "loss": 2.9281,
+      "step": 18020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001937077182976262,
+      "loss": 2.91,
+      "step": 18030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001933627357182894,
+      "loss": 2.9021,
+      "step": 18040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019301775313895257,
+      "loss": 2.9014,
+      "step": 18050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019267277055961575,
+      "loss": 2.9067,
+      "step": 18060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019232778798027893,
+      "loss": 2.9107,
+      "step": 18070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001919828054009421,
+      "loss": 2.9041,
+      "step": 18080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001916378228216053,
+      "loss": 2.9034,
+      "step": 18090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019129284024226845,
+      "loss": 2.899,
+      "step": 18100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019094785766293163,
+      "loss": 2.9059,
+      "step": 18110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001906028750835948,
+      "loss": 2.9003,
+      "step": 18120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000190257892504258,
+      "loss": 2.9058,
+      "step": 18130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018991290992492114,
+      "loss": 2.9004,
+      "step": 18140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018956792734558433,
+      "loss": 2.9004,
+      "step": 18150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001892229447662475,
+      "loss": 2.9072,
+      "step": 18160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001888779621869107,
+      "loss": 2.8917,
+      "step": 18170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018853297960757387,
+      "loss": 2.91,
+      "step": 18180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018818799702823705,
+      "loss": 2.9152,
+      "step": 18190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018784301444890023,
+      "loss": 2.9106,
+      "step": 18200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001874980318695634,
+      "loss": 2.9006,
+      "step": 18210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001871530492902266,
+      "loss": 2.9132,
+      "step": 18220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018680806671088975,
+      "loss": 2.9076,
+      "step": 18230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018646308413155293,
+      "loss": 2.9111,
+      "step": 18240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001861181015522161,
+      "loss": 2.8961,
+      "step": 18250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001857731189728793,
+      "loss": 2.904,
+      "step": 18260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018542813639354244,
+      "loss": 2.8988,
+      "step": 18270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018508315381420562,
+      "loss": 2.8957,
+      "step": 18280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001847381712348688,
+      "loss": 2.8993,
+      "step": 18290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000184393188655532,
+      "loss": 2.9022,
+      "step": 18300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018404820607619517,
+      "loss": 2.9161,
+      "step": 18310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018370322349685835,
+      "loss": 2.908,
+      "step": 18320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018335824091752153,
+      "loss": 2.9033,
+      "step": 18330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001830132583381847,
+      "loss": 2.9005,
+      "step": 18340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001826682757588479,
+      "loss": 2.8951,
+      "step": 18350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018232329317951105,
+      "loss": 2.9066,
+      "step": 18360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018197831060017423,
+      "loss": 2.8975,
+      "step": 18370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001816333280208374,
+      "loss": 2.8977,
+      "step": 18380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001812883454415006,
+      "loss": 2.9046,
+      "step": 18390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018094336286216374,
+      "loss": 2.9123,
+      "step": 18400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018059838028282692,
+      "loss": 2.8857,
+      "step": 18410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001802533977034901,
+      "loss": 2.9016,
+      "step": 18420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017990841512415329,
+      "loss": 2.8923,
+      "step": 18430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017956343254481647,
+      "loss": 2.9027,
+      "step": 18440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017921844996547965,
+      "loss": 2.908,
+      "step": 18450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017887346738614283,
+      "loss": 2.9026,
+      "step": 18460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000178528484806806,
+      "loss": 2.9011,
+      "step": 18470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001781835022274692,
+      "loss": 2.9134,
+      "step": 18480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017783851964813235,
+      "loss": 2.9094,
+      "step": 18490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017749353706879553,
+      "loss": 2.9168,
+      "step": 18500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001771485544894587,
+      "loss": 2.9078,
+      "step": 18510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001768035719101219,
+      "loss": 2.9088,
+      "step": 18520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017645858933078504,
+      "loss": 2.8929,
+      "step": 18530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017611360675144822,
+      "loss": 2.9097,
+      "step": 18540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001757686241721114,
+      "loss": 2.9171,
+      "step": 18550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017542364159277459,
+      "loss": 2.9058,
+      "step": 18560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017507865901343777,
+      "loss": 2.9026,
+      "step": 18570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017473367643410095,
+      "loss": 2.911,
+      "step": 18580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017438869385476413,
+      "loss": 2.9003,
+      "step": 18590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001740437112754273,
+      "loss": 2.8921,
+      "step": 18600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001736987286960905,
+      "loss": 2.8978,
+      "step": 18610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017335374611675365,
+      "loss": 2.903,
+      "step": 18620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017300876353741683,
+      "loss": 2.9167,
+      "step": 18630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017266378095808,
+      "loss": 2.9008,
+      "step": 18640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001723187983787432,
+      "loss": 2.9158,
+      "step": 18650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017197381579940634,
+      "loss": 2.8964,
+      "step": 18660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017162883322006952,
+      "loss": 2.9017,
+      "step": 18670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001712838506407327,
+      "loss": 2.8938,
+      "step": 18680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017093886806139589,
+      "loss": 2.8902,
+      "step": 18690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017059388548205907,
+      "loss": 2.9042,
+      "step": 18700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017024890290272225,
+      "loss": 2.908,
+      "step": 18710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016990392032338543,
+      "loss": 2.9169,
+      "step": 18720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001695589377440486,
+      "loss": 2.9048,
+      "step": 18730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001692139551647118,
+      "loss": 2.897,
+      "step": 18740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016886897258537494,
+      "loss": 2.8963,
+      "step": 18750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016852399000603813,
+      "loss": 2.8995,
+      "step": 18760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001681790074267013,
+      "loss": 2.9097,
+      "step": 18770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001678340248473645,
+      "loss": 2.9062,
+      "step": 18780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016748904226802764,
+      "loss": 2.9075,
+      "step": 18790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016714405968869082,
+      "loss": 2.898,
+      "step": 18800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000166799077109354,
+      "loss": 2.8948,
+      "step": 18810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016645409453001719,
+      "loss": 2.9047,
+      "step": 18820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016610911195068037,
+      "loss": 2.9082,
+      "step": 18830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016576412937134355,
+      "loss": 2.9023,
+      "step": 18840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016541914679200673,
+      "loss": 2.8907,
+      "step": 18850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001650741642126699,
+      "loss": 2.8995,
+      "step": 18860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001647291816333331,
+      "loss": 2.9033,
+      "step": 18870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016438419905399624,
+      "loss": 2.8937,
+      "step": 18880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016403921647465943,
+      "loss": 2.9065,
+      "step": 18890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001636942338953226,
+      "loss": 2.9081,
+      "step": 18900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001633492513159858,
+      "loss": 2.8908,
+      "step": 18910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016300426873664894,
+      "loss": 2.8854,
+      "step": 18920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016265928615731212,
+      "loss": 2.9197,
+      "step": 18930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001623143035779753,
+      "loss": 2.9064,
+      "step": 18940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016196932099863848,
+      "loss": 2.9003,
+      "step": 18950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016162433841930167,
+      "loss": 2.9129,
+      "step": 18960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016127935583996482,
+      "loss": 2.897,
+      "step": 18970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016093437326062803,
+      "loss": 2.8948,
+      "step": 18980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001605893906812912,
+      "loss": 2.9096,
+      "step": 18990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001602444081019544,
+      "loss": 2.8969,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015989942552261754,
+      "loss": 2.9047,
+      "step": 19010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00015955444294328072,
+      "loss": 2.9074,
+      "step": 19020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001592094603639439,
+      "loss": 2.8973,
+      "step": 19030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001588644777846071,
+      "loss": 2.8889,
+      "step": 19040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015851949520527024,
+      "loss": 2.9068,
+      "step": 19050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015817451262593342,
+      "loss": 2.8999,
+      "step": 19060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001578295300465966,
+      "loss": 2.9058,
+      "step": 19070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015748454746725978,
+      "loss": 2.907,
+      "step": 19080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015713956488792297,
+      "loss": 2.8975,
+      "step": 19090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015679458230858612,
+      "loss": 2.9147,
+      "step": 19100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001564495997292493,
+      "loss": 2.8968,
+      "step": 19110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015610461714991248,
+      "loss": 2.9069,
+      "step": 19120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001557596345705757,
+      "loss": 2.9031,
+      "step": 19130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015541465199123884,
+      "loss": 2.9043,
+      "step": 19140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015506966941190202,
+      "loss": 2.8963,
+      "step": 19150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001547246868325652,
+      "loss": 2.9008,
+      "step": 19160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001543797042532284,
+      "loss": 2.9275,
+      "step": 19170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015403472167389154,
+      "loss": 2.9009,
+      "step": 19180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015368973909455472,
+      "loss": 2.9053,
+      "step": 19190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001533447565152179,
+      "loss": 2.8904,
+      "step": 19200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015299977393588108,
+      "loss": 2.8968,
+      "step": 19210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015265479135654426,
+      "loss": 2.9082,
+      "step": 19220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015230980877720742,
+      "loss": 2.8934,
+      "step": 19230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001519648261978706,
+      "loss": 2.9071,
+      "step": 19240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015161984361853378,
+      "loss": 2.9159,
+      "step": 19250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015127486103919696,
+      "loss": 2.9136,
+      "step": 19260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00015092987845986014,
+      "loss": 2.9004,
+      "step": 19270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015058489588052332,
+      "loss": 2.8979,
+      "step": 19280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001502399133011865,
+      "loss": 2.8922,
+      "step": 19290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014989493072184969,
+      "loss": 2.8993,
+      "step": 19300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014954994814251284,
+      "loss": 2.9041,
+      "step": 19310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014920496556317602,
+      "loss": 2.8833,
+      "step": 19320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001488599829838392,
+      "loss": 2.8982,
+      "step": 19330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014851500040450238,
+      "loss": 2.9048,
+      "step": 19340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014817001782516556,
+      "loss": 2.8915,
+      "step": 19350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014782503524582872,
+      "loss": 2.8908,
+      "step": 19360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001474800526664919,
+      "loss": 2.9058,
+      "step": 19370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014713507008715508,
+      "loss": 2.8945,
+      "step": 19380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014679008750781826,
+      "loss": 2.8981,
+      "step": 19390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014644510492848144,
+      "loss": 2.9122,
+      "step": 19400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014610012234914462,
+      "loss": 2.8942,
+      "step": 19410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001457551397698078,
+      "loss": 2.9072,
+      "step": 19420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014541015719047099,
+      "loss": 2.8965,
+      "step": 19430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014506517461113414,
+      "loss": 2.8834,
+      "step": 19440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014472019203179732,
+      "loss": 2.9076,
+      "step": 19450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001443752094524605,
+      "loss": 2.8834,
+      "step": 19460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014403022687312368,
+      "loss": 2.8852,
+      "step": 19470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014368524429378686,
+      "loss": 2.8897,
+      "step": 19480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014334026171445002,
+      "loss": 2.8979,
+      "step": 19490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001429952791351132,
+      "loss": 2.891,
+      "step": 19500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014265029655577638,
+      "loss": 2.8932,
+      "step": 19510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014230531397643956,
+      "loss": 2.9052,
+      "step": 19520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014196033139710274,
+      "loss": 2.9028,
+      "step": 19530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014161534881776592,
+      "loss": 2.8781,
+      "step": 19540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001412703662384291,
+      "loss": 2.8978,
+      "step": 19550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014092538365909229,
+      "loss": 2.8984,
+      "step": 19560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014058040107975544,
+      "loss": 2.9055,
+      "step": 19570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014023541850041862,
+      "loss": 2.8859,
+      "step": 19580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001398904359210818,
+      "loss": 2.9003,
+      "step": 19590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013954545334174498,
+      "loss": 2.8944,
+      "step": 19600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013920047076240816,
+      "loss": 2.8844,
+      "step": 19610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013885548818307132,
+      "loss": 2.9048,
+      "step": 19620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001385105056037345,
+      "loss": 2.9139,
+      "step": 19630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013816552302439768,
+      "loss": 2.9128,
+      "step": 19640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013782054044506086,
+      "loss": 2.897,
+      "step": 19650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013747555786572404,
+      "loss": 2.9038,
+      "step": 19660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013713057528638722,
+      "loss": 2.9033,
+      "step": 19670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001367855927070504,
+      "loss": 2.8934,
+      "step": 19680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013644061012771358,
+      "loss": 2.8899,
+      "step": 19690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013609562754837674,
+      "loss": 2.8918,
+      "step": 19700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013575064496903992,
+      "loss": 2.9018,
+      "step": 19710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001354056623897031,
+      "loss": 2.8867,
+      "step": 19720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013506067981036628,
+      "loss": 2.8971,
+      "step": 19730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013471569723102946,
+      "loss": 2.8968,
+      "step": 19740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013437071465169262,
+      "loss": 2.9017,
+      "step": 19750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001340257320723558,
+      "loss": 2.9061,
+      "step": 19760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013368074949301898,
+      "loss": 2.895,
+      "step": 19770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013333576691368216,
+      "loss": 2.9108,
+      "step": 19780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013299078433434534,
+      "loss": 2.907,
+      "step": 19790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013264580175500852,
+      "loss": 2.8903,
+      "step": 19800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001323008191756717,
+      "loss": 2.8924,
+      "step": 19810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013195583659633488,
+      "loss": 2.9111,
+      "step": 19820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013161085401699804,
+      "loss": 2.9126,
+      "step": 19830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013126587143766122,
+      "loss": 2.9008,
+      "step": 19840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001309208888583244,
+      "loss": 2.8989,
+      "step": 19850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013057590627898758,
+      "loss": 2.9063,
+      "step": 19860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013023092369965076,
+      "loss": 2.8996,
+      "step": 19870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012988594112031392,
+      "loss": 2.8897,
+      "step": 19880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001295409585409771,
+      "loss": 2.8948,
+      "step": 19890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012919597596164028,
+      "loss": 2.8946,
+      "step": 19900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012885099338230346,
+      "loss": 2.9119,
+      "step": 19910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012850601080296664,
+      "loss": 2.9039,
+      "step": 19920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012816102822362982,
+      "loss": 2.903,
+      "step": 19930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000127816045644293,
+      "loss": 2.8977,
+      "step": 19940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012747106306495618,
+      "loss": 2.8685,
+      "step": 19950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012712608048561934,
+      "loss": 2.8942,
+      "step": 19960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012678109790628252,
+      "loss": 2.8754,
+      "step": 19970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001264361153269457,
+      "loss": 2.9161,
+      "step": 19980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012609113274760888,
+      "loss": 2.8968,
+      "step": 19990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012574615016827206,
+      "loss": 2.885,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012540116758893522,
+      "loss": 2.8934,
+      "step": 20010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001250561850095984,
+      "loss": 2.8749,
+      "step": 20020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012471120243026158,
+      "loss": 2.8922,
+      "step": 20030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012436621985092476,
+      "loss": 2.9168,
+      "step": 20040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001240212372715879,
+      "loss": 2.9134,
+      "step": 20050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012367625469225112,
+      "loss": 2.9046,
+      "step": 20060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001233312721129143,
+      "loss": 2.8867,
+      "step": 20070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012298628953357748,
+      "loss": 2.9099,
+      "step": 20080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012264130695424064,
+      "loss": 2.8851,
+      "step": 20090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012229632437490382,
+      "loss": 2.8799,
+      "step": 20100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000121951341795567,
+      "loss": 2.8917,
+      "step": 20110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012160635921623018,
+      "loss": 2.9111,
+      "step": 20120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012126137663689336,
+      "loss": 2.8918,
+      "step": 20130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012091639405755652,
+      "loss": 2.8854,
+      "step": 20140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001205714114782197,
+      "loss": 2.9046,
+      "step": 20150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012022642889888288,
+      "loss": 2.8943,
+      "step": 20160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011988144631954606,
+      "loss": 2.9105,
+      "step": 20170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011953646374020923,
+      "loss": 2.8898,
+      "step": 20180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011919148116087241,
+      "loss": 2.887,
+      "step": 20190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011884649858153559,
+      "loss": 2.8793,
+      "step": 20200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011850151600219877,
+      "loss": 2.8991,
+      "step": 20210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011815653342286192,
+      "loss": 2.9083,
+      "step": 20220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011781155084352512,
+      "loss": 2.8901,
+      "step": 20230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001174665682641883,
+      "loss": 2.9144,
+      "step": 20240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011712158568485148,
+      "loss": 2.9012,
+      "step": 20250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011677660310551466,
+      "loss": 2.9078,
+      "step": 20260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011643162052617782,
+      "loss": 2.8972,
+      "step": 20270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000116086637946841,
+      "loss": 2.9002,
+      "step": 20280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011574165536750418,
+      "loss": 2.9027,
+      "step": 20290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011539667278816736,
+      "loss": 2.8845,
+      "step": 20300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011505169020883053,
+      "loss": 2.9042,
+      "step": 20310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011470670762949371,
+      "loss": 2.8946,
+      "step": 20320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011436172505015689,
+      "loss": 2.8914,
+      "step": 20330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011401674247082007,
+      "loss": 2.9043,
+      "step": 20340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011367175989148322,
+      "loss": 2.8961,
+      "step": 20350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001133267773121464,
+      "loss": 2.8782,
+      "step": 20360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001129817947328096,
+      "loss": 2.8724,
+      "step": 20370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011263681215347278,
+      "loss": 2.8965,
+      "step": 20380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011229182957413596,
+      "loss": 2.8866,
+      "step": 20390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011194684699479911,
+      "loss": 2.8989,
+      "step": 20400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001116018644154623,
+      "loss": 2.9017,
+      "step": 20410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011125688183612548,
+      "loss": 2.9098,
+      "step": 20420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011091189925678866,
+      "loss": 2.8817,
+      "step": 20430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011056691667745183,
+      "loss": 2.9081,
+      "step": 20440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000110221934098115,
+      "loss": 2.886,
+      "step": 20450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010987695151877819,
+      "loss": 2.8952,
+      "step": 20460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010953196893944137,
+      "loss": 2.8998,
+      "step": 20470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010918698636010452,
+      "loss": 2.8699,
+      "step": 20480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001088420037807677,
+      "loss": 2.8835,
+      "step": 20490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010849702120143088,
+      "loss": 2.9056,
+      "step": 20500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010815203862209408,
+      "loss": 2.8813,
+      "step": 20510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010780705604275726,
+      "loss": 2.9017,
+      "step": 20520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010746207346342041,
+      "loss": 2.8973,
+      "step": 20530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001071170908840836,
+      "loss": 2.8789,
+      "step": 20540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010677210830474678,
+      "loss": 2.887,
+      "step": 20550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010642712572540996,
+      "loss": 2.8994,
+      "step": 20560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010608214314607313,
+      "loss": 2.8967,
+      "step": 20570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001057371605667363,
+      "loss": 2.8931,
+      "step": 20580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010539217798739949,
+      "loss": 2.8972,
+      "step": 20590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010504719540806267,
+      "loss": 2.902,
+      "step": 20600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010470221282872582,
+      "loss": 2.8832,
+      "step": 20610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000104357230249389,
+      "loss": 2.8945,
+      "step": 20620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010401224767005218,
+      "loss": 2.9078,
+      "step": 20630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010366726509071537,
+      "loss": 2.893,
+      "step": 20640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010332228251137856,
+      "loss": 2.8932,
+      "step": 20650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010297729993204171,
+      "loss": 2.8875,
+      "step": 20660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001026323173527049,
+      "loss": 2.8974,
+      "step": 20670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010228733477336808,
+      "loss": 2.8849,
+      "step": 20680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010194235219403126,
+      "loss": 2.903,
+      "step": 20690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010159736961469442,
+      "loss": 2.8959,
+      "step": 20700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001012523870353576,
+      "loss": 2.8957,
+      "step": 20710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010090740445602079,
+      "loss": 2.889,
+      "step": 20720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010056242187668397,
+      "loss": 2.9019,
+      "step": 20730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010021743929734712,
+      "loss": 2.9055,
+      "step": 20740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.98724567180103e-05,
+      "loss": 2.9057,
+      "step": 20750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.952747413867348e-05,
+      "loss": 2.8997,
+      "step": 20760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.918249155933666e-05,
+      "loss": 2.895,
+      "step": 20770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.883750897999985e-05,
+      "loss": 2.892,
+      "step": 20780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.849252640066303e-05,
+      "loss": 2.9127,
+      "step": 20790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.81475438213262e-05,
+      "loss": 2.9084,
+      "step": 20800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.780256124198938e-05,
+      "loss": 2.8953,
+      "step": 20810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.745757866265254e-05,
+      "loss": 2.9005,
+      "step": 20820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.711259608331572e-05,
+      "loss": 2.9115,
+      "step": 20830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.67676135039789e-05,
+      "loss": 2.9035,
+      "step": 20840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.642263092464209e-05,
+      "loss": 2.8946,
+      "step": 20850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.607764834530525e-05,
+      "loss": 2.8842,
+      "step": 20860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.573266576596843e-05,
+      "loss": 2.9001,
+      "step": 20870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.53876831866316e-05,
+      "loss": 2.9008,
+      "step": 20880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.504270060729478e-05,
+      "loss": 2.8893,
+      "step": 20890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.469771802795796e-05,
+      "loss": 2.9054,
+      "step": 20900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.435273544862115e-05,
+      "loss": 2.8966,
+      "step": 20910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.400775286928433e-05,
+      "loss": 2.9063,
+      "step": 20920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.36627702899475e-05,
+      "loss": 2.8839,
+      "step": 20930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.331778771061068e-05,
+      "loss": 2.8948,
+      "step": 20940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.297280513127384e-05,
+      "loss": 2.9006,
+      "step": 20950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.262782255193702e-05,
+      "loss": 2.8956,
+      "step": 20960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.228283997260019e-05,
+      "loss": 2.8925,
+      "step": 20970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.193785739326339e-05,
+      "loss": 2.9124,
+      "step": 20980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.159287481392655e-05,
+      "loss": 2.8778,
+      "step": 20990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.124789223458973e-05,
+      "loss": 2.8999,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.090290965525292e-05,
+      "loss": 2.8932,
+      "step": 21010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.055792707591608e-05,
+      "loss": 2.9099,
+      "step": 21020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.021294449657926e-05,
+      "loss": 2.9047,
+      "step": 21030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.986796191724243e-05,
+      "loss": 2.8765,
+      "step": 21040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.952297933790563e-05,
+      "loss": 2.8945,
+      "step": 21050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.91779967585688e-05,
+      "loss": 2.9207,
+      "step": 21060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.883301417923197e-05,
+      "loss": 2.8947,
+      "step": 21070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.848803159989514e-05,
+      "loss": 2.8758,
+      "step": 21080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.814304902055832e-05,
+      "loss": 2.8955,
+      "step": 21090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.779806644122149e-05,
+      "loss": 2.8748,
+      "step": 21100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.745308386188467e-05,
+      "loss": 2.9001,
+      "step": 21110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.710810128254785e-05,
+      "loss": 2.8904,
+      "step": 21120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.676311870321103e-05,
+      "loss": 2.9,
+      "step": 21130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.641813612387422e-05,
+      "loss": 2.8717,
+      "step": 21140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.607315354453738e-05,
+      "loss": 2.8916,
+      "step": 21150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.572817096520056e-05,
+      "loss": 2.9036,
+      "step": 21160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.538318838586373e-05,
+      "loss": 2.8819,
+      "step": 21170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.503820580652691e-05,
+      "loss": 2.8866,
+      "step": 21180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.46932232271901e-05,
+      "loss": 2.9136,
+      "step": 21190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.434824064785327e-05,
+      "loss": 2.9073,
+      "step": 21200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.400325806851644e-05,
+      "loss": 2.8884,
+      "step": 21210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.365827548917962e-05,
+      "loss": 2.8842,
+      "step": 21220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.331329290984279e-05,
+      "loss": 2.8965,
+      "step": 21230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.296831033050597e-05,
+      "loss": 2.8846,
+      "step": 21240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.262332775116915e-05,
+      "loss": 2.885,
+      "step": 21250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.227834517183233e-05,
+      "loss": 2.8824,
+      "step": 21260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.193336259249551e-05,
+      "loss": 2.8736,
+      "step": 21270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.158838001315868e-05,
+      "loss": 2.8761,
+      "step": 21280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.124339743382186e-05,
+      "loss": 2.8939,
+      "step": 21290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.089841485448503e-05,
+      "loss": 2.8697,
+      "step": 21300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.055343227514821e-05,
+      "loss": 2.8798,
+      "step": 21310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.020844969581139e-05,
+      "loss": 2.8794,
+      "step": 21320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.986346711647457e-05,
+      "loss": 2.8797,
+      "step": 21330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.951848453713774e-05,
+      "loss": 2.8832,
+      "step": 21340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.917350195780092e-05,
+      "loss": 2.8838,
+      "step": 21350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.882851937846409e-05,
+      "loss": 2.8892,
+      "step": 21360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.848353679912727e-05,
+      "loss": 2.8686,
+      "step": 21370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.813855421979045e-05,
+      "loss": 2.9019,
+      "step": 21380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.779357164045363e-05,
+      "loss": 2.8639,
+      "step": 21390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.744858906111681e-05,
+      "loss": 2.883,
+      "step": 21400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.710360648177998e-05,
+      "loss": 2.8911,
+      "step": 21410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.675862390244316e-05,
+      "loss": 2.8875,
+      "step": 21420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.641364132310633e-05,
+      "loss": 2.8933,
+      "step": 21430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.606865874376951e-05,
+      "loss": 2.901,
+      "step": 21440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.572367616443269e-05,
+      "loss": 2.8648,
+      "step": 21450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.537869358509587e-05,
+      "loss": 2.8826,
+      "step": 21460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.503371100575904e-05,
+      "loss": 2.8957,
+      "step": 21470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.468872842642222e-05,
+      "loss": 2.8748,
+      "step": 21480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.434374584708539e-05,
+      "loss": 2.8684,
+      "step": 21490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.399876326774857e-05,
+      "loss": 2.8666,
+      "step": 21500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.365378068841174e-05,
+      "loss": 2.8934,
+      "step": 21510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.330879810907493e-05,
+      "loss": 2.8934,
+      "step": 21520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.296381552973811e-05,
+      "loss": 2.8861,
+      "step": 21530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.261883295040128e-05,
+      "loss": 2.8991,
+      "step": 21540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.227385037106446e-05,
+      "loss": 2.8901,
+      "step": 21550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.192886779172763e-05,
+      "loss": 2.8848,
+      "step": 21560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.158388521239081e-05,
+      "loss": 2.8795,
+      "step": 21570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.123890263305398e-05,
+      "loss": 2.8936,
+      "step": 21580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.089392005371717e-05,
+      "loss": 2.8726,
+      "step": 21590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.054893747438034e-05,
+      "loss": 2.889,
+      "step": 21600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.020395489504352e-05,
+      "loss": 2.8838,
+      "step": 21610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.985897231570669e-05,
+      "loss": 2.8798,
+      "step": 21620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.951398973636987e-05,
+      "loss": 2.881,
+      "step": 21630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.916900715703304e-05,
+      "loss": 2.8937,
+      "step": 21640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.882402457769622e-05,
+      "loss": 2.8934,
+      "step": 21650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.847904199835941e-05,
+      "loss": 2.8796,
+      "step": 21660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.813405941902258e-05,
+      "loss": 2.8852,
+      "step": 21670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.778907683968576e-05,
+      "loss": 2.8777,
+      "step": 21680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.744409426034893e-05,
+      "loss": 2.8872,
+      "step": 21690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.709911168101211e-05,
+      "loss": 2.8884,
+      "step": 21700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.675412910167528e-05,
+      "loss": 2.8722,
+      "step": 21710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.640914652233846e-05,
+      "loss": 2.8818,
+      "step": 21720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.606416394300164e-05,
+      "loss": 2.8876,
+      "step": 21730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.571918136366482e-05,
+      "loss": 2.8981,
+      "step": 21740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.537419878432799e-05,
+      "loss": 2.8944,
+      "step": 21750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.502921620499117e-05,
+      "loss": 2.8859,
+      "step": 21760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.468423362565434e-05,
+      "loss": 2.8864,
+      "step": 21770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.433925104631752e-05,
+      "loss": 2.8836,
+      "step": 21780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.39942684669807e-05,
+      "loss": 2.8951,
+      "step": 21790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.364928588764388e-05,
+      "loss": 2.8803,
+      "step": 21800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.330430330830706e-05,
+      "loss": 2.876,
+      "step": 21810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.295932072897023e-05,
+      "loss": 2.8884,
+      "step": 21820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.261433814963341e-05,
+      "loss": 2.8872,
+      "step": 21830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.226935557029658e-05,
+      "loss": 2.8848,
+      "step": 21840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.192437299095976e-05,
+      "loss": 2.8849,
+      "step": 21850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.157939041162294e-05,
+      "loss": 2.888,
+      "step": 21860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.123440783228612e-05,
+      "loss": 2.8789,
+      "step": 21870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.088942525294929e-05,
+      "loss": 2.8888,
+      "step": 21880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.054444267361247e-05,
+      "loss": 2.8734,
+      "step": 21890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.0199460094275644e-05,
+      "loss": 2.8537,
+      "step": 21900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9854477514938825e-05,
+      "loss": 2.859,
+      "step": 21910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9509494935602006e-05,
+      "loss": 2.8833,
+      "step": 21920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.916451235626517e-05,
+      "loss": 2.8861,
+      "step": 21930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8819529776928354e-05,
+      "loss": 2.8931,
+      "step": 21940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.847454719759153e-05,
+      "loss": 2.8905,
+      "step": 21950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.812956461825471e-05,
+      "loss": 2.8845,
+      "step": 21960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7784582038917884e-05,
+      "loss": 2.8635,
+      "step": 21970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.7439599459581065e-05,
+      "loss": 2.8863,
+      "step": 21980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.709461688024423e-05,
+      "loss": 2.8641,
+      "step": 21990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6749634300907414e-05,
+      "loss": 2.8733,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.640465172157059e-05,
+      "loss": 2.8847,
+      "step": 22010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.605966914223377e-05,
+      "loss": 2.8722,
+      "step": 22020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5714686562896936e-05,
+      "loss": 2.8901,
+      "step": 22030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5369703983560124e-05,
+      "loss": 2.8858,
+      "step": 22040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5024721404223305e-05,
+      "loss": 2.8546,
+      "step": 22050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.467973882488647e-05,
+      "loss": 2.8914,
+      "step": 22060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.4334756245549654e-05,
+      "loss": 2.897,
+      "step": 22070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.398977366621283e-05,
+      "loss": 2.8669,
+      "step": 22080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.364479108687601e-05,
+      "loss": 2.9008,
+      "step": 22090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.329980850753918e-05,
+      "loss": 2.89,
+      "step": 22100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2954825928202365e-05,
+      "loss": 2.8717,
+      "step": 22110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.260984334886553e-05,
+      "loss": 2.8887,
+      "step": 22120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.226486076952871e-05,
+      "loss": 2.8901,
+      "step": 22130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.191987819019189e-05,
+      "loss": 2.8822,
+      "step": 22140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.157489561085507e-05,
+      "loss": 2.865,
+      "step": 22150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.1229913031518236e-05,
+      "loss": 2.8767,
+      "step": 22160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.088493045218142e-05,
+      "loss": 2.898,
+      "step": 22170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.05399478728446e-05,
+      "loss": 2.877,
+      "step": 22180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.019496529350777e-05,
+      "loss": 2.8856,
+      "step": 22190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.984998271417095e-05,
+      "loss": 2.8727,
+      "step": 22200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.950500013483413e-05,
+      "loss": 2.8525,
+      "step": 22210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.91600175554973e-05,
+      "loss": 2.8837,
+      "step": 22220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8815034976160476e-05,
+      "loss": 2.8878,
+      "step": 22230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847005239682366e-05,
+      "loss": 2.8873,
+      "step": 22240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.812506981748684e-05,
+      "loss": 2.8729,
+      "step": 22250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.778008723815001e-05,
+      "loss": 2.8893,
+      "step": 22260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.743510465881319e-05,
+      "loss": 2.8849,
+      "step": 22270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.709012207947637e-05,
+      "loss": 2.8758,
+      "step": 22280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.674513950013954e-05,
+      "loss": 2.8727,
+      "step": 22290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.640015692080272e-05,
+      "loss": 2.877,
+      "step": 22300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.60551743414659e-05,
+      "loss": 2.881,
+      "step": 22310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.571019176212907e-05,
+      "loss": 2.8934,
+      "step": 22320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5365209182792246e-05,
+      "loss": 2.8738,
+      "step": 22330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.502022660345542e-05,
+      "loss": 2.8989,
+      "step": 22340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.46752440241186e-05,
+      "loss": 2.8873,
+      "step": 22350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.433026144478178e-05,
+      "loss": 2.8818,
+      "step": 22360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.398527886544496e-05,
+      "loss": 2.8916,
+      "step": 22370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.364029628610814e-05,
+      "loss": 2.8864,
+      "step": 22380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.329531370677131e-05,
+      "loss": 2.8792,
+      "step": 22390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2950331127434487e-05,
+      "loss": 2.8811,
+      "step": 22400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.260534854809766e-05,
+      "loss": 2.8952,
+      "step": 22410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.226036596876084e-05,
+      "loss": 2.8931,
+      "step": 22420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1915383389424016e-05,
+      "loss": 2.8605,
+      "step": 22430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.157040081008719e-05,
+      "loss": 2.8874,
+      "step": 22440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.122541823075037e-05,
+      "loss": 2.9019,
+      "step": 22450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0880435651413546e-05,
+      "loss": 2.8867,
+      "step": 22460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.053545307207672e-05,
+      "loss": 2.8863,
+      "step": 22470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.01904704927399e-05,
+      "loss": 2.8827,
+      "step": 22480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.984548791340308e-05,
+      "loss": 2.8743,
+      "step": 22490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9500505334066257e-05,
+      "loss": 2.8885,
+      "step": 22500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.915552275472943e-05,
+      "loss": 2.8732,
+      "step": 22510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.881054017539261e-05,
+      "loss": 2.8874,
+      "step": 22520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8465557596055786e-05,
+      "loss": 2.8819,
+      "step": 22530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.812057501671896e-05,
+      "loss": 2.8828,
+      "step": 22540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.777559243738214e-05,
+      "loss": 2.8768,
+      "step": 22550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7430609858045316e-05,
+      "loss": 2.9018,
+      "step": 22560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.708562727870849e-05,
+      "loss": 2.8767,
+      "step": 22570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.674064469937167e-05,
+      "loss": 2.883,
+      "step": 22580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6395662120034845e-05,
+      "loss": 2.897,
+      "step": 22590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.605067954069802e-05,
+      "loss": 2.8813,
+      "step": 22600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5705696961361194e-05,
+      "loss": 2.883,
+      "step": 22610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.536071438202438e-05,
+      "loss": 2.8757,
+      "step": 22620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5015731802687556e-05,
+      "loss": 2.8888,
+      "step": 22630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.467074922335073e-05,
+      "loss": 2.8827,
+      "step": 22640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.432576664401391e-05,
+      "loss": 2.8961,
+      "step": 22650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3980784064677086e-05,
+      "loss": 2.8952,
+      "step": 22660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.363580148534026e-05,
+      "loss": 2.8763,
+      "step": 22670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3290818906003434e-05,
+      "loss": 2.89,
+      "step": 22680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2945836326666615e-05,
+      "loss": 2.8793,
+      "step": 22690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.260085374732979e-05,
+      "loss": 2.8967,
+      "step": 22700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2255871167992964e-05,
+      "loss": 2.9027,
+      "step": 22710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1910888588656145e-05,
+      "loss": 2.8841,
+      "step": 22720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.156590600931932e-05,
+      "loss": 2.8717,
+      "step": 22730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1220923429982494e-05,
+      "loss": 2.8851,
+      "step": 22740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0875940850645675e-05,
+      "loss": 2.8724,
+      "step": 22750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.0530958271308856e-05,
+      "loss": 2.8686,
+      "step": 22760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.018597569197203e-05,
+      "loss": 2.8936,
+      "step": 22770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9840993112635208e-05,
+      "loss": 2.8773,
+      "step": 22780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9496010533298382e-05,
+      "loss": 2.8862,
+      "step": 22790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.915102795396156e-05,
+      "loss": 2.8743,
+      "step": 22800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8806045374624737e-05,
+      "loss": 2.884,
+      "step": 22810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.846106279528791e-05,
+      "loss": 2.8794,
+      "step": 22820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.811608021595109e-05,
+      "loss": 2.8789,
+      "step": 22830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7771097636614263e-05,
+      "loss": 2.8738,
+      "step": 22840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.742611505727744e-05,
+      "loss": 2.8858,
+      "step": 22850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.708113247794062e-05,
+      "loss": 2.8593,
+      "step": 22860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6736149898603793e-05,
+      "loss": 2.888,
+      "step": 22870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6391167319266978e-05,
+      "loss": 2.8772,
+      "step": 22880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6046184739930152e-05,
+      "loss": 2.8706,
+      "step": 22890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.570120216059333e-05,
+      "loss": 2.8742,
+      "step": 22900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5356219581256504e-05,
+      "loss": 2.8825,
+      "step": 22910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.501123700191968e-05,
+      "loss": 2.8676,
+      "step": 22920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.466625442258286e-05,
+      "loss": 2.8822,
+      "step": 22930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4321271843246033e-05,
+      "loss": 2.8873,
+      "step": 22940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.397628926390921e-05,
+      "loss": 2.8964,
+      "step": 22950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.363130668457239e-05,
+      "loss": 2.8948,
+      "step": 22960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3286324105235566e-05,
+      "loss": 2.8894,
+      "step": 22970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2941341525898744e-05,
+      "loss": 2.8873,
+      "step": 22980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.259635894656192e-05,
+      "loss": 2.8849,
+      "step": 22990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2251376367225096e-05,
+      "loss": 2.8716,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.190639378788827e-05,
+      "loss": 2.8807,
+      "step": 23010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1561411208551448e-05,
+      "loss": 2.9003,
+      "step": 23020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1216428629214626e-05,
+      "loss": 2.8768,
+      "step": 23030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0871446049877803e-05,
+      "loss": 2.8542,
+      "step": 23040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052646347054098e-05,
+      "loss": 2.876,
+      "step": 23050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0181480891204155e-05,
+      "loss": 2.8816,
+      "step": 23060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9836498311867333e-05,
+      "loss": 2.8713,
+      "step": 23070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.949151573253051e-05,
+      "loss": 2.8912,
+      "step": 23080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9146533153193685e-05,
+      "loss": 2.8887,
+      "step": 23090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8801550573856866e-05,
+      "loss": 2.8801,
+      "step": 23100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.845656799452004e-05,
+      "loss": 2.882,
+      "step": 23110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8111585415183218e-05,
+      "loss": 2.8825,
+      "step": 23120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7766602835846396e-05,
+      "loss": 2.8711,
+      "step": 23130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.742162025650957e-05,
+      "loss": 2.8795,
+      "step": 23140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7076637677172748e-05,
+      "loss": 2.8795,
+      "step": 23150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6731655097835922e-05,
+      "loss": 2.8796,
+      "step": 23160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6386672518499103e-05,
+      "loss": 2.8858,
+      "step": 23170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6041689939162277e-05,
+      "loss": 2.8858,
+      "step": 23180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696707359825455e-05,
+      "loss": 2.8636,
+      "step": 23190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5351724780488633e-05,
+      "loss": 2.8869,
+      "step": 23200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5006742201151809e-05,
+      "loss": 2.8761,
+      "step": 23210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4661759621814985e-05,
+      "loss": 2.8874,
+      "step": 23220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4316777042478164e-05,
+      "loss": 2.8801,
+      "step": 23230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.397179446314134e-05,
+      "loss": 2.8876,
+      "step": 23240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3626811883804516e-05,
+      "loss": 2.877,
+      "step": 23250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3281829304467692e-05,
+      "loss": 2.8764,
+      "step": 23260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.293684672513087e-05,
+      "loss": 2.8705,
+      "step": 23270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2591864145794045e-05,
+      "loss": 2.8745,
+      "step": 23280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2246881566457223e-05,
+      "loss": 2.8824,
+      "step": 23290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1901898987120399e-05,
+      "loss": 2.8843,
+      "step": 23300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1556916407783577e-05,
+      "loss": 2.8975,
+      "step": 23310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1211933828446754e-05,
+      "loss": 2.8828,
+      "step": 23320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.086695124910993e-05,
+      "loss": 2.8706,
+      "step": 23330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0521968669773108e-05,
+      "loss": 2.8661,
+      "step": 23340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0176986090436284e-05,
+      "loss": 2.8569,
+      "step": 23350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.83200351109946e-06,
+      "loss": 2.8808,
+      "step": 23360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.487020931762638e-06,
+      "loss": 2.8857,
+      "step": 23370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.142038352425815e-06,
+      "loss": 2.8708,
+      "step": 23380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.797055773088991e-06,
+      "loss": 2.8993,
+      "step": 23390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.452073193752167e-06,
+      "loss": 2.8934,
+      "step": 23400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.107090614415345e-06,
+      "loss": 2.886,
+      "step": 23410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.762108035078521e-06,
+      "loss": 2.8669,
+      "step": 23420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.417125455741698e-06,
+      "loss": 2.8937,
+      "step": 23430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0721428764048755e-06,
+      "loss": 2.8805,
+      "step": 23440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.727160297068052e-06,
+      "loss": 2.8577,
+      "step": 23450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.382177717731228e-06,
+      "loss": 2.8796,
+      "step": 23460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.037195138394406e-06,
+      "loss": 2.8786,
+      "step": 23470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.692212559057583e-06,
+      "loss": 2.8833,
+      "step": 23480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.347229979720759e-06,
+      "loss": 2.8809,
+      "step": 23490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0022474003839365e-06,
+      "loss": 2.8823,
+      "step": 23500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.657264821047113e-06,
+      "loss": 2.8975,
+      "step": 23510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.312282241710289e-06,
+      "loss": 2.8713,
+      "step": 23520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.967299662373467e-06,
+      "loss": 2.9029,
+      "step": 23530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.622317083036643e-06,
+      "loss": 2.8656,
+      "step": 23540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.27733450369982e-06,
+      "loss": 2.8755,
+      "step": 23550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.932351924362997e-06,
+      "loss": 2.8946,
+      "step": 23560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.587369345026174e-06,
+      "loss": 2.8941,
+      "step": 23570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2423867656893506e-06,
+      "loss": 2.8879,
+      "step": 23580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8974041863525275e-06,
+      "loss": 2.8741,
+      "step": 23590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524216070157043e-06,
+      "loss": 2.8808,
+      "step": 23600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2074390276788811e-06,
+      "loss": 2.8844,
+      "step": 23610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.62456448342058e-07,
+      "loss": 2.8866,
+      "step": 23620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.174738690052349e-07,
+      "loss": 2.8767,
+      "step": 23630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.724912896684116e-07,
+      "loss": 2.8647,
+      "step": 23640
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 2.882770538330078,
+      "eval_runtime": 2001.771,
+      "eval_samples_per_second": 11.812,
+      "eval_steps_per_second": 2.953,
+      "step": 23645
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 23645,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 8.607098951691878e+16,
+  "trial_name": null,
+  "trial_params": null
+}