File size: 237,192 Bytes
78aa4ee
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Igc5itf-xMGj"
   },
   "source": [
    "# Masakhane - Machine Translation for African Languages (Using JoeyNMT)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "x4fXCKCf36IK"
   },
   "source": [
    "## Note before beginning:\n",
    "### - The idea is that you should be able to make minimal changes to this in order to get SOME result for your own translation corpus. \n",
    "\n",
    "### - The tl;dr: Go to the **\"TODO\"** comments which will tell you what to update to get up and running\n",
    "\n",
    "### - If you actually want to have a clue what you're doing, read the text and peek at the links\n",
    "\n",
    "### - With 100 epochs, it should take around 7 hours to run in Google Colab\n",
    "\n",
    "### - Once you've gotten a result for your language, please attach and email your notebook that generated it to [email protected]\n",
    "\n",
    "### - If you care enough and get a chance, doing a brief background on your language would be amazing. See examples in  [(Martinus, 2019)](https://arxiv.org/abs/1906.05685)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "l929HimrxS0a"
   },
   "source": [
    "## Retrieve your data & make a parallel corpus\n",
    "\n",
    "If you are wanting to use the JW300 data referenced on the Masakhane website or in our GitHub repo, you can use `opus-tools` to convert the data into a convenient format. `opus_read` from that package provides a convenient tool for reading the native aligned XML files and to convert them to TMX format. The tool can also be used to fetch relevant files from OPUS on the fly and to filter the data as necessary. [Read the documentation](https://pypi.org/project/opustools-pkg/) for more details.\n",
    "\n",
    "Once you have your corpus files in TMX format (an xml structure which will include the sentences in your target language and your source language in a single file), we recommend reading them into a pandas dataframe. Thankfully, Jade wrote a silly `tmx2dataframe` package which converts your tmx file to a pandas dataframe. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "oGRmDELn7Az0"
   },
   "outputs": [],
   "source": [
    "# from google.colab import drive\n",
    "# drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "Cn3tgQLzUxwn"
   },
   "outputs": [],
   "source": [
    "# TODO: Set your source and target languages. Keep in mind, these traditionally use language codes as found here:\n",
    "# These will also become the suffix's of all vocab and corpus files used throughout\n",
    "import os\n",
    "source_language = \"en\"\n",
    "target_language = \"luo\" \n",
    "lc = False  # If True, lowercase the data.\n",
    "seed = 42  # Random seed for shuffling.\n",
    "tag = \"baseline\" # Give a unique name to your folder - this is to ensure you don't rewrite any models you've already submitted\n",
    "\n",
    "os.environ[\"src\"] = source_language # Sets them in bash as well, since we often use bash scripts\n",
    "os.environ[\"tgt\"] = target_language\n",
    "os.environ[\"tag\"] = tag\n",
    "\n",
    "# This will save it to a folder in our gdrive instead!\n",
    "# !mkdir -p \"/content/drive/My Drive/masakhane/$src-$tgt-$tag\"\n",
    "# os.environ[\"gdrive_path\"] = \"/content/drive/My Drive/masakhane/%s-%s-%s\" % (source_language, target_language, tag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "kBSgJHEw7Nvx"
   },
   "outputs": [],
   "source": [
    "# !echo $gdrive_path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "gA75Fs9ys8Y9"
   },
   "outputs": [],
   "source": [
    "# Install opus-tools\n",
    "#! pip install opustools-pkg"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Uncomment cell below if notebook is being run for the first time and you need to download the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xq-tDZVks7ZD"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Alignment file /proj/nlpl/data/OPUS/JW300/latest/xml/en-luo.xml.gz not found. The following files are available for downloading:\n",
      "\n",
      "   1 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/en-luo.xml.gz\n",
      " 263 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/en.zip\n",
      "  14 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/luo.zip\n",
      "\n",
      " 279 MB Total size\n",
      "./JW300_latest_xml_en-luo.xml.gz ... 100% of 1 MB\n",
      "./JW300_latest_xml_en.zip ... 100% of 263 MB\n",
      "./JW300_latest_xml_luo.zip ... 100% of 14 MB\n"
     ]
    }
   ],
   "source": [
    "# Downloading our corpus\n",
    "! opus_read -d JW300 -s $src -t $tgt -wm moses -w jw300.$src jw300.$tgt -q\n",
    "\n",
    "# extract the corpus file\n",
    "! gunzip JW300_latest_xml_$src-$tgt.xml.gz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "n48GDRnP8y2G"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2020-02-19 06:24:03--  https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-any.en\n",
      "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.192.133, 151.101.128.133, 151.101.64.133, ...\n",
      "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.192.133|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 277791 (271K) [text/plain]\n",
      "Saving to: ‘test.en-any.en’\n",
      "\n",
      "test.en-any.en      100%[===================>] 271.28K  --.-KB/s    in 0.04s   \n",
      "\n",
      "2020-02-19 06:24:04 (6.13 MB/s) - ‘test.en-any.en’ saved [277791/277791]\n",
      "\n",
      "--2020-02-19 06:24:04--  https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-luo.en\n",
      "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.192.133, 151.101.128.133, 151.101.64.133, ...\n",
      "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.192.133|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 205584 (201K) [text/plain]\n",
      "Saving to: ‘test.en-luo.en’\n",
      "\n",
      "test.en-luo.en      100%[===================>] 200.77K  --.-KB/s    in 0.04s   \n",
      "\n",
      "2020-02-19 06:24:04 (5.24 MB/s) - ‘test.en-luo.en’ saved [205584/205584]\n",
      "\n",
      "--2020-02-19 06:24:04--  https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-luo.luo\n",
      "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.192.133, 151.101.128.133, 151.101.64.133, ...\n",
      "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.192.133|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 209877 (205K) [text/plain]\n",
      "Saving to: ‘test.en-luo.luo’\n",
      "\n",
      "test.en-luo.luo     100%[===================>] 204.96K  --.-KB/s    in 0.04s   \n",
      "\n",
      "2020-02-19 06:24:05 (4.95 MB/s) - ‘test.en-luo.luo’ saved [209877/209877]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Download the global test set.\n",
    "! wget https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-any.en\n",
    "  \n",
    "# And the specific test set for this language pair.\n",
    "os.environ[\"trg\"] = target_language \n",
    "os.environ[\"src\"] = source_language \n",
    "\n",
    "! wget https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-$trg.en \n",
    "! mv test.en-$trg.en test.en\n",
    "! wget https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-$trg.$trg \n",
    "! mv test.en-$trg.$trg test.$trg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "NqDG-CI28y2L"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 3571 global test sentences to filter from the training/dev data.\n"
     ]
    }
   ],
   "source": [
    "# Read the test data to filter from train and dev splits.\n",
    "# Store english portion in set for quick filtering checks.\n",
    "en_test_sents = set()\n",
    "filter_test_sents = \"test.en-any.en\"\n",
    "j = 0\n",
    "with open(filter_test_sents) as f:\n",
    "  for line in f:\n",
    "    en_test_sents.add(line.strip())\n",
    "    j += 1\n",
    "print('Loaded {} global test sentences to filter from the training/dev data.'.format(j))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "3CNdwLBCfSIl"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded data and skipped 4654/154941 lines since contained in test set.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>source_sentence</th>\n",
       "      <th>target_sentence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Compassion in a Cruel World</td>\n",
       "      <td>Kecho Ji e Piny ma Ji Ok Dew Jowetegi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>A MAN in Burundi falls seriously ill with mala...</td>\n",
       "      <td>MALARIA mager ogoyo dichwo moro e piny Burundi .</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>He urgently needs to be transferred to a hospi...</td>\n",
       "      <td>Dwarore otere e osiptal mapiyo ahinya .</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     source_sentence  \\\n",
       "0                        Compassion in a Cruel World   \n",
       "1  A MAN in Burundi falls seriously ill with mala...   \n",
       "2  He urgently needs to be transferred to a hospi...   \n",
       "\n",
       "                                    target_sentence  \n",
       "0             Kecho Ji e Piny ma Ji Ok Dew Jowetegi  \n",
       "1  MALARIA mager ogoyo dichwo moro e piny Burundi .  \n",
       "2           Dwarore otere e osiptal mapiyo ahinya .  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "# TMX file to dataframe\n",
    "source_file = 'jw300.' + source_language\n",
    "target_file = 'jw300.' + target_language\n",
    "\n",
    "source = []\n",
    "target = []\n",
    "skip_lines = []  # Collect the line numbers of the source portion to skip the same lines for the target portion.\n",
    "with open(source_file) as f:\n",
    "    for i, line in enumerate(f):\n",
    "        # Skip sentences that are contained in the test set.\n",
    "        if line.strip() not in en_test_sents:\n",
    "            source.append(line.strip())\n",
    "        else:\n",
    "            skip_lines.append(i)             \n",
    "with open(target_file) as f:\n",
    "    for j, line in enumerate(f):\n",
    "        # Only add to corpus if corresponding source was not skipped.\n",
    "        if j not in skip_lines:\n",
    "            target.append(line.strip())\n",
    "    \n",
    "print('Loaded data and skipped {}/{} lines since contained in test set.'.format(len(skip_lines), i))\n",
    "    \n",
    "df = pd.DataFrame(zip(source, target), columns=['source_sentence', 'target_sentence'])\n",
    "df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "YkuK3B4p2AkN"
   },
   "source": [
    "## Pre-processing and export\n",
    "\n",
    "It is generally a good idea to remove duplicate translations and conflicting translations from the corpus. In practice, these public corpora include some number of these that need to be cleaned.\n",
    "\n",
    "In addition we will split our data into dev/test/train and export to the filesystem."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "M_2ouEOH1_1q"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:8: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:9: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  if __name__ == '__main__':\n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:12: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  if sys.path[0] == '':\n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:13: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  del sys.path[0]\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "# drop duplicate translations\n",
    "df_pp = df.drop_duplicates()\n",
    "\n",
    "#drop empty lines (alp)\n",
    "df_pp['source_sentence'].replace('', np.nan, inplace=True)\n",
    "df_pp['target_sentence'].replace('', np.nan, inplace=True)\n",
    "df_pp.dropna(subset=['source_sentence'], inplace=True)\n",
    "df_pp.dropna(subset=['target_sentence'], inplace=True)\n",
    "\n",
    "# drop conflicting translations\n",
    "df_pp.drop_duplicates(subset='source_sentence', inplace=True)\n",
    "df_pp.drop_duplicates(subset='target_sentence', inplace=True)\n",
    "\n",
    "# Shuffle the data to remove bias in dev set selection.\n",
    "df_pp = df_pp.sample(frac=1, random_state=seed).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "hxxBOCA-xXhy"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==> train.bpe.en <==\n",
      "Paul wrote : “ Let the su@@ n not set with you in a pro@@ v@@ ok@@ ed st@@ ate . ”\n",
      "Some of these en@@ emi@@ es as@@ sa@@ ul@@ t Jehovah’s Witnesses physi@@ c@@ ally .\n",
      "S@@ T@@ U@@ D@@ Y A@@ R@@ TI@@ C@@ L@@ E@@ S 3 , 4 PA@@ G@@ E@@ S 19 - 28\n",
      "From inf@@ an@@ cy , children can be@@ gin lear@@ ning about Jehovah’s Word .\n",
      "In these ways , youn@@ ger ones and those ne@@ w@@ ly associ@@ ated learn to per@@ form ac@@ ts of kindness for others .\n",
      "O@@ f@@ fered S@@ el@@ ves in W@@ est A@@ fri@@ c@@ a , 1 / 15\n",
      "Those resurrec@@ ted to heaven will even@@ tually number 14@@ 4@@ ,000 .\n",
      "A@@ ga@@ inst the bac@@ k@@ d@@ ro@@ p of Satan’s dis@@ as@@ tr@@ ous rule , Jehovah’s perfect qualities are even more ob@@ vi@@ ous than they o@@ ther@@ wise might have been .\n",
      "The first Kingdom H@@ all in B@@ a@@ uru ​ — a r@@ ented place with a sig@@ n I pa@@ inted , 195@@ 5\n",
      "In ad@@ dition , his s@@ mok@@ ing end@@ ang@@ ers the health of those around him .\n",
      "\n",
      "==> train.bpe.luo <==\n",
      "Paulo nondiko kama : “ Kik uwe chieng ’ p@@ odhi ka pod un gi mirima . ”\n",
      "Moko kuom jo@@ wasi@@ gu@@ gi goyo Joneno mag Jehova e yo ma@@ sie .\n",
      "S@@ U@@ L@@ A MA@@ G P@@ U@@ ON@@ J@@ R@@ UO@@ K 3 , 4 IT@@ E MA@@ G 19 - 28\n",
      "A e tin - gi , nyithindo nyalo chako puonjore Wach Jehova .\n",
      "Ka jodongo otiyo kanyachiel gi rowere kod joma nyi@@ en e konyo jomoko , gin bende gibiro bedo gi chuny mar timo kamano .\n",
      "Pinyruodh Nyasaye No@@ chako Lo@@ cho K@@ ar@@ ang’o ?\n",
      "Jogo duto mi@@ biro chi@@ ero mondo odhi e polo kwan - gi en 14@@ 4@@ ,000 .\n",
      "Nikech wanyalo neno kaka loch Satan en loch manono , kido malong’o chuth mag Jehova koro nyis@@ or@@ en@@ wa e yo maler moloyo .\n",
      "O@@ d Romo ma ne wak@@ wongo bedogo e taon mar B@@ a@@ uru , en ot ma ne wa@@ p@@ ango ma no@@ ke@@ tie ranyisi ma n@@ an@@ diko , 195@@ 5\n",
      "E wi mano , ma@@ dh@@ one n@@ da@@ wa bende ketho ngima jogo man machiegni kode .\n",
      "\n",
      "==> train.en <==\n",
      "Paul wrote : “ Let the sun not set with you in a provoked state . ”\n",
      "Some of these enemies assault Jehovah’s Witnesses physically .\n",
      "STUDY ARTICLES 3 , 4 PAGES 19 - 28\n",
      "From infancy , children can begin learning about Jehovah’s Word .\n",
      "In these ways , younger ones and those newly associated learn to perform acts of kindness for others .\n",
      "Offered Selves in West Africa , 1 / 15\n",
      "Those resurrected to heaven will eventually number 144,000 .\n",
      "Against the backdrop of Satan’s disastrous rule , Jehovah’s perfect qualities are even more obvious than they otherwise might have been .\n",
      "The first Kingdom Hall in Bauru ​ — a rented place with a sign I painted , 1955\n",
      "In addition , his smoking endangers the health of those around him .\n",
      "\n",
      "==> train.luo <==\n",
      "Paulo nondiko kama : “ Kik uwe chieng ’ podhi ka pod un gi mirima . ”\n",
      "Moko kuom jowasigugi goyo Joneno mag Jehova e yo masie .\n",
      "SULA MAG PUONJRUOK 3 , 4 ITE MAG 19 - 28\n",
      "A e tin - gi , nyithindo nyalo chako puonjore Wach Jehova .\n",
      "Ka jodongo otiyo kanyachiel gi rowere kod joma nyien e konyo jomoko , gin bende gibiro bedo gi chuny mar timo kamano .\n",
      "Pinyruodh Nyasaye Nochako Locho Karang’o ?\n",
      "Jogo duto mibiro chiero mondo odhi e polo kwan - gi en 144,000 .\n",
      "Nikech wanyalo neno kaka loch Satan en loch manono , kido malong’o chuth mag Jehova koro nyisorenwa e yo maler moloyo .\n",
      "Od Romo ma ne wakwongo bedogo e taon mar Bauru , en ot ma ne wapango ma noketie ranyisi ma nandiko , 1955\n",
      "E wi mano , madhone ndawa bende ketho ngima jogo man machiegni kode .\n",
      "==> dev.bpe.en <==\n",
      "Their own willing obedi@@ ence to what Jehovah requ@@ ires should demonstr@@ ate that God’s rul@@ es are reason@@ able .\n",
      "It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "Se@@ e@@ ing ourselves in the light of what the Scriptures say can in@@ deed be a powerful a@@ id in identi@@ f@@ ying the inten@@ tions of the heart .\n",
      "He had written about the “ rest@@ or@@ ation of all things of which God spoke through the m@@ outh of his holy prophe@@ ts of old time . ”\n",
      "15 O@@ U@@ R R@@ E@@ A@@ D@@ ER@@ S A@@ S@@ K . . .\n",
      "A@@ M@@ ON@@ G the ze@@ al@@ ous Witnesses serving in l@@ ands where there is a greater need for more Kingdom pre@@ ach@@ ers are num@@ erous single sisters .\n",
      "But often they are not .\n",
      "As our Creator , Jehovah deser@@ ves pra@@ ise , glory , than@@ k@@ s@@ giving , and honor .\n",
      "An@@ cient King Solomon had a lo@@ t of knowledge about Jehovah’s ways .\n",
      "The writ@@ er of Psalm 1@@ 11 had the cor@@ rec@@ t view of godly fear , st@@ ating : “ The fear of Jehovah is the begin@@ ning of wisdom . All those doing them [ keep@@ ing his comman@@ d@@ ments ] have a good in@@ sight . ” ​ — Ps .\n",
      "\n",
      "==> dev.bpe.luo <==\n",
      "Ka gin giwegi gi@@ ikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo mak@@ gi .\n",
      "Mano bende miyo am@@ edo miye luor kaka wuon@@ wa kendo kaka ng’at mat@@ ay@@ owa e weche mag lamo . ”\n",
      "Ka wan@@ on@@ ore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "No@@ yudo D@@ un@@ n osen@@ diko wach “ loso gik moko duto od@@ ok makare , kaka Nyasaye nowacho gi dho j@@ on@@ abi duto maler nyaka a chak@@ ruok piny . ”\n",
      "15 J@@ O@@ S@@ O@@ M@@ B@@ W@@ A P@@ EN@@ J@@ O . . .\n",
      "N@@ I@@ TI@@ E nyimine mang’eny ma pok oken@@ di mo@@ chiwore tiyo ne Jehova gi kinda e pinje ma jolendo dwar@@ or@@ ee ahinya .\n",
      "To ok tim@@ re@@ ga kamano .\n",
      "Nikech Jehova en Ja@@ chu@@ e@@ ch@@ wa , onego wa@@ miye duong ’ , wa@@ g@@ one erokamano , kendo wa@@ miye luor matut .\n",
      "Ruoth Solomon nong’eyo gik mathoth e wi Jehova .\n",
      "Jal ma ne ondiko Zaburi 1@@ 11 ne nigi paro makare kuom wach luoro Nyasaye kane owacho kama : “ Lu@@ oro Jehova en chak@@ r@@ w@@ ok mar rieko ; jo ma timo kamano [ ma@@ rito chi@@ kene ] weche don@@ j@@ oni@@ gi maber . ” ​ ​ —⁠ Zab .\n",
      "\n",
      "==> dev.en <==\n",
      "Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "15 OUR READERS ASK . . .\n",
      "AMONG the zealous Witnesses serving in lands where there is a greater need for more Kingdom preachers are numerous single sisters .\n",
      "But often they are not .\n",
      "As our Creator , Jehovah deserves praise , glory , thanksgiving , and honor .\n",
      "Ancient King Solomon had a lot of knowledge about Jehovah’s ways .\n",
      "The writer of Psalm 111 had the correct view of godly fear , stating : “ The fear of Jehovah is the beginning of wisdom . All those doing them [ keeping his commandments ] have a good insight . ” ​ — Ps .\n",
      "\n",
      "==> dev.luo <==\n",
      "Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "15 JOSOMBWA PENJO . . .\n",
      "NITIE nyimine mang’eny ma pok okendi mochiwore tiyo ne Jehova gi kinda e pinje ma jolendo dwaroree ahinya .\n",
      "To ok timrega kamano .\n",
      "Nikech Jehova en Jachuechwa , onego wamiye duong ’ , wagone erokamano , kendo wamiye luor matut .\n",
      "Ruoth Solomon nong’eyo gik mathoth e wi Jehova .\n",
      "Jal ma ne ondiko Zaburi 111 ne nigi paro makare kuom wach luoro Nyasaye kane owacho kama : “ Luoro Jehova en chakrwok mar rieko ; jo ma timo kamano [ marito chikene ] weche donjonigi maber . ” ​ ​ —⁠ Zab .\n"
     ]
    }
   ],
   "source": [
    "# This section does the split between train/dev for the parallel corpora then saves them as separate files\n",
    "# We use 1000 dev test and the given test set.\n",
    "import csv\n",
    "\n",
    "# Do the split between dev/train and create parallel corpora\n",
    "num_dev_patterns = 1000\n",
    "\n",
    "# Optional: lower case the corpora - this will make it easier to generalize, but without proper casing.\n",
    "if lc:  # Julia: making lowercasing optional\n",
    "    df_pp[\"source_sentence\"] = df_pp[\"source_sentence\"].str.lower()\n",
    "    df_pp[\"target_sentence\"] = df_pp[\"target_sentence\"].str.lower()\n",
    "\n",
    "# Julia: test sets are already generated\n",
    "dev = df_pp.tail(num_dev_patterns) # Herman: Error in original\n",
    "stripped = df_pp.drop(df_pp.tail(num_dev_patterns).index)\n",
    "\n",
    "with open(\"train.\"+source_language, \"w\") as src_file, open(\"train.\"+target_language, \"w\") as trg_file:\n",
    "  for index, row in stripped.iterrows():\n",
    "    src_file.write(row[\"source_sentence\"]+\"\\n\")\n",
    "    trg_file.write(row[\"target_sentence\"]+\"\\n\")\n",
    "    \n",
    "with open(\"dev.\"+source_language, \"w\") as src_file, open(\"dev.\"+target_language, \"w\") as trg_file:\n",
    "  for index, row in dev.iterrows():\n",
    "    src_file.write(row[\"source_sentence\"]+\"\\n\")\n",
    "    trg_file.write(row[\"target_sentence\"]+\"\\n\")\n",
    "\n",
    "#stripped[[\"source_sentence\"]].to_csv(\"train.\"+source_language, header=False, index=False)  # Herman: Added `header=False` everywhere\n",
    "#stripped[[\"target_sentence\"]].to_csv(\"train.\"+target_language, header=False, index=False)  # Julia: Problematic handling of quotation marks.\n",
    "\n",
    "#dev[[\"source_sentence\"]].to_csv(\"dev.\"+source_language, header=False, index=False)\n",
    "#dev[[\"target_sentence\"]].to_csv(\"dev.\"+target_language, header=False, index=False)\n",
    "\n",
    "\n",
    "# Doublecheck the format below. There should be no extra quotation marks or weird characters.\n",
    "! head train.*\n",
    "! head dev.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "epeCydmCyS8X"
   },
   "source": [
    "\n",
    "\n",
    "---\n",
    "\n",
    "\n",
    "## Installation of JoeyNMT\n",
    "\n",
    "JoeyNMT is a simple, minimalist NMT package which is useful for learning and teaching. Check out the documentation for JoeyNMT [here](https://joeynmt.readthedocs.io)  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "iBRMm4kMxZ8L"
   },
   "outputs": [],
   "source": [
    "# Install JoeyNMT\n",
    "#! git clone https://github.com/joeynmt/joeynmt.git\n",
    "#! cd joeynmt; pip3 install ."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "AaE77Tcppex9"
   },
   "source": [
    "# Preprocessing the Data into Subword BPE Tokens\n",
    "\n",
    "- One of the most powerful improvements for agglutinative languages (a feature of most Bantu languages) is using BPE tokenization [ (Sennrich, 2015) ](https://arxiv.org/abs/1508.07909).\n",
    "\n",
    "- It was also shown that by optimizing the umber of BPE codes we significantly improve results for low-resourced languages [(Sennrich, 2019)](https://www.aclweb.org/anthology/P19-1021) [(Martinus, 2019)](https://arxiv.org/abs/1906.05685)\n",
    "\n",
    "- Below we have the scripts for doing BPE tokenization of our data. We use 4000 tokens as recommended by [(Sennrich, 2019)](https://www.aclweb.org/anthology/P19-1021). You do not need to change anything. Simply running the below will be suitable. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# One of the huge boosts in NMT performance was to use a different method of tokenizing. \n",
    "# Usually, NMT would tokenize by words. However, using a method called BPE gave amazing boosts to performance\n",
    "\n",
    "# Do subword NMT\n",
    "from os import path\n",
    "os.environ[\"src\"] = source_language # Sets them in bash as well, since we often use bash scripts\n",
    "os.environ[\"tgt\"] = target_language\n",
    "\n",
    "# Learn BPEs on the training data.\n",
    "os.environ[\"data_path\"] = path.join(\"../../joeynmt\", \"data\", source_language + target_language) # Herman! \n",
    "! subword-nmt learn-joint-bpe-and-vocab --input train.$src train.$tgt -s 4000 -o bpe.codes.4000 --write-vocabulary vocab.$src vocab.$tgt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Apply BPE splits to the development and test data.\n",
    "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$src < train.$src > train.bpe.$src\n",
    "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$tgt < train.$tgt > train.bpe.$tgt\n",
    "\n",
    "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$src < dev.$src > dev.bpe.$src\n",
    "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$tgt < dev.$tgt > dev.bpe.$tgt\n",
    "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$src < test.$src > test.bpe.$src\n",
    "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$tgt < test.$tgt > test.bpe.$tgt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "! sudo chmod 777 ../../joeynmt/data/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "bpe.codes.4000\tdev.en\t     test.bpe.luo    test.luo\t    train.en\r\n",
      "dev.bpe.en\tdev.luo      test.en\t     train.bpe.en   train.luo\r\n",
      "dev.bpe.luo\ttest.bpe.en  test.en-any.en  train.bpe.luo  vocab.txt\r\n"
     ]
    }
   ],
   "source": [
    "# Create directory, move everyone we care about to the correct location\n",
    "! mkdir -p $data_path\n",
    "! cp train.* $data_path\n",
    "! cp test.* $data_path\n",
    "! cp dev.* $data_path\n",
    "! cp bpe.codes.4000 $data_path\n",
    "! ls $data_path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Also move everything we care about to a mounted location in google drive (relevant if running in colab) at gdrive_path\n",
    "# ! cp train.* \"$gdrive_path\"\n",
    "# ! cp test.* \"$gdrive_path\"\n",
    "# ! cp dev.* \"$gdrive_path\"\n",
    "# ! cp bpe.codes.4000 \"$gdrive_path\"\n",
    "# ! ls \"$gdrive_path\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Create that vocab using build_vocab\n",
    "! sudo chmod 777 ../../joeynmt/scripts/build_vocab.py\n",
    "! ../../joeynmt/scripts/build_vocab.py ../../joeynmt/data/$src$tgt/train.bpe.$src ../../joeynmt/data/$src$tgt/train.bpe.$tgt --output_path ../../joeynmt/data/$src$tgt/vocab.txt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "H-TyjtmXB1mL"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BPE Luo Sentences\n",
      "O@@ k@@ um@@ ba malach mar yie ( Ne paragraf mar 12 - 14 )\n",
      "O@@ g@@ ud@@ u mar war@@ ruok ( Ne paragraf mar 15 - 18 )\n",
      "A@@ se@@ fwenyo ni ji chiko it@@ gi sama gineno ni i@@ hero weche manie Muma , kendo itimo duto ma inyalo mondo i@@ kony@@ gi . ”\n",
      "L@@ ig@@ ang@@ la mar roho maler ( Ne paragraf mar 19 - 20 )\n",
      "Kata kamano , kokalo kuom teko mar Jehova wanyalo k@@ wede !\n",
      "Combined BPE Vocab\n",
      "portunity\n",
      "aburi\n",
      "Q\n",
      ".E\n",
      "Ç@@\n",
      "Ł@@\n",
      "ʺ\n",
      "rans@@\n",
      "oura@@\n",
      "_\n"
     ]
    }
   ],
   "source": [
    "# Some output\n",
    "! echo \"BPE Luo Sentences\"\n",
    "! tail -n 5 test.bpe.$tgt\n",
    "! echo \"Combined BPE Vocab\"\n",
    "! tail -n 10 ../../joeynmt/data/$src$tgt/vocab.txt  # Herman"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "IlMitUHR8Qy-"
   },
   "outputs": [],
   "source": [
    "# Also move everything we care about to a mounted location in google drive (relevant if running in colab) at gdrive_path\n",
    "#! cp train.* \"$gdrive_path\"\n",
    "#! cp test.* \"$gdrive_path\"\n",
    "#! cp dev.* \"$gdrive_path\"\n",
    "#! cp bpe.codes.4000 \"$gdrive_path\"\n",
    "#! ls \"$gdrive_path\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ixmzi60WsUZ8"
   },
   "source": [
    "# Creating the JoeyNMT Config\n",
    "\n",
    "JoeyNMT requires a yaml config. We provide a template below. We've also set a number of defaults with it, that you may play with!\n",
    "\n",
    "- We used Transformer architecture \n",
    "- We set our dropout to reasonably high: 0.3 (recommended in  [(Sennrich, 2019)](https://www.aclweb.org/anthology/P19-1021))\n",
    "\n",
    "Things worth playing with:\n",
    "- The batch size (also recommended to change for low-resourced languages)\n",
    "- The number of epochs (we've set it at 30 just so it runs in about an hour, for testing purposes)\n",
    "- The decoder options (beam_size, alpha)\n",
    "- Evaluation metrics (BLEU versus Crhf4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "PIs1lY2hxMsl"
   },
   "outputs": [],
   "source": [
    "# This creates the config file for our JoeyNMT system. It might seem overwhelming so we've provided a couple of useful parameters you'll need to update\n",
    "# (You can of course play with all the parameters if you'd like!)\n",
    "\n",
    "name = '%s%s' % (source_language, target_language)\n",
    "# gdrive_path = os.environ[\"gdrive_path\"]\n",
    "\n",
    "# Create the config\n",
    "config = \"\"\"\n",
    "name: \"{name}_transformer\"\n",
    "\n",
    "data:\n",
    "    src: \"{source_language}\"\n",
    "    trg: \"{target_language}\"\n",
    "    train: \"data/{name}/train.bpe\"\n",
    "    dev:   \"data/{name}/dev.bpe\"\n",
    "    test:  \"data/{name}/test.bpe\"\n",
    "    level: \"bpe\"\n",
    "    lowercase: False\n",
    "    max_sent_length: 100\n",
    "    src_vocab: \"data/{name}/vocab.txt\"\n",
    "    trg_vocab: \"data/{name}/vocab.txt\"\n",
    "\n",
    "testing:\n",
    "    beam_size: 5\n",
    "    alpha: 1.0\n",
    "\n",
    "training:\n",
    "    #load_model: \"models/{name}_transformer/1000.ckpt\" # if uncommented, load a pre-trained model from this checkpoint\n",
    "    random_seed: 42\n",
    "    optimizer: \"adam\"\n",
    "    normalization: \"tokens\"\n",
    "    adam_betas: [0.9, 0.999] \n",
    "    scheduling: \"plateau\"           # TODO: try switching from plateau to Noam scheduling\n",
    "    patience: 5                     # For plateau: decrease learning rate by decrease_factor if validation score has not improved for this many validation rounds.\n",
    "    learning_rate_factor: 0.5       # factor for Noam scheduler (used with Transformer)\n",
    "    learning_rate_warmup: 1000      # warmup steps for Noam scheduler (used with Transformer)\n",
    "    decrease_factor: 0.7\n",
    "    loss: \"crossentropy\"\n",
    "    learning_rate: 0.0003\n",
    "    learning_rate_min: 0.00000001\n",
    "    weight_decay: 0.0\n",
    "    label_smoothing: 0.1\n",
    "    batch_size: 4096\n",
    "    batch_type: \"token\"\n",
    "    eval_batch_size: 3600\n",
    "    eval_batch_type: \"token\"\n",
    "    batch_multiplier: 1\n",
    "    early_stopping_metric: \"ppl\"\n",
    "    epochs: 30                     # TODO: Decrease for when playing around and checking of working. Around 30 is sufficient to check if its working at all\n",
    "    validation_freq: 1000          # TODO: Set to at least once per epoch.\n",
    "    logging_freq: 100\n",
    "    eval_metric: \"bleu\"\n",
    "    model_dir: \"models/{name}_transformer\"\n",
    "    overwrite: False               # TODO: Set to True if you want to overwrite possibly existing models. \n",
    "    shuffle: True\n",
    "    use_cuda: True\n",
    "    max_output_length: 100\n",
    "    print_valid_sents: [0, 1, 2, 3]\n",
    "    keep_last_ckpts: 3\n",
    "\n",
    "model:\n",
    "    initializer: \"xavier\"\n",
    "    bias_initializer: \"zeros\"\n",
    "    init_gain: 1.0\n",
    "    embed_initializer: \"xavier\"\n",
    "    embed_init_gain: 1.0\n",
    "    tied_embeddings: True\n",
    "    tied_softmax: True\n",
    "    encoder:\n",
    "        type: \"transformer\"\n",
    "        num_layers: 6\n",
    "        num_heads: 4             # TODO: Increase to 8 for larger data.\n",
    "        embeddings:\n",
    "            embedding_dim: 256   # TODO: Increase to 512 for larger data.\n",
    "            scale: True\n",
    "            dropout: 0.2\n",
    "        # typically ff_size = 4 x hidden_size\n",
    "        hidden_size: 256         # TODO: Increase to 512 for larger data.\n",
    "        ff_size: 1024            # TODO: Increase to 2048 for larger data.\n",
    "        dropout: 0.3\n",
    "    decoder:\n",
    "        type: \"transformer\"\n",
    "        num_layers: 6\n",
    "        num_heads: 4              # TODO: Increase to 8 for larger data.\n",
    "        embeddings:\n",
    "            embedding_dim: 256    # TODO: Increase to 512 for larger data.\n",
    "            scale: True\n",
    "            dropout: 0.2\n",
    "        # typically ff_size = 4 x hidden_size\n",
    "        hidden_size: 256         # TODO: Increase to 512 for larger data.\n",
    "        ff_size: 1024            # TODO: Increase to 2048 for larger data.\n",
    "        dropout: 0.3\n",
    "\"\"\".format(name=name, gdrive_path=\"n/a\", source_language=source_language, target_language=target_language)\n",
    "with open(\"../../joeynmt/configs/transformer_{name}.yaml\".format(name=name),'w') as f:\n",
    "    f.write(config)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "pIifxE3Qzuvs"
   },
   "source": [
    "# Train the Model\n",
    "\n",
    "This single line of joeynmt runs the training using the config we made above"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total 136\r\n",
      "drwxr-xr-x 11 root root  4096 Oct 25 12:22 .\r\n",
      "drwxrwxrwx 11 root root  4096 Feb 19 06:15 ..\r\n",
      "drwxr-xr-x  8 root root  4096 Oct 24 15:14 .git\r\n",
      "-rw-r--r--  1 root root    49 Oct 24 15:14 .gitattributes\r\n",
      "drwxr-xr-x  3 root root  4096 Oct 24 15:14 .github\r\n",
      "-rw-r--r--  1 root root    71 Oct 24 15:14 .gitignore\r\n",
      "-rw-r--r--  1 root root 13514 Oct 24 15:14 .pylintrc\r\n",
      "-rw-r--r--  1 root root   159 Oct 24 15:14 .readthedocs.yml\r\n",
      "-rw-r--r--  1 root root   542 Oct 24 15:14 .travis.yml\r\n",
      "-rwxrw-rwx  1 root root  3354 Oct 24 15:14 CODE_OF_CONDUCT.md\r\n",
      "-rwxrw-rwx  1 root root  1071 Oct 24 15:14 LICENSE\r\n",
      "-rwxrw-rwx  1 root root 13286 Oct 24 15:14 README.md\r\n",
      "-rwxrw-rwx  1 root root  8229 Oct 24 15:14 benchmarks.md\r\n",
      "drwxrw-rwx  3 root root  4096 Feb 19 06:30 configs\r\n",
      "drwxrwxrwx  8 root root  4096 Feb 19 06:35 data\r\n",
      "drwxrw-rwx  4 root root  4096 Oct 24 15:14 docs\r\n",
      "-rwxrw-rwx  1 root root 14373 Oct 24 15:14 joey-small.png\r\n",
      "drwxrw-rwx  3 root root  4096 Oct 24 16:35 joeynmt\r\n",
      "drwxrwxrwx  7 root root  4096 Feb 20 07:38 models\r\n",
      "-rwxrw-rwx  1 root root   167 Oct 24 15:14 requirements.txt\r\n",
      "drwxrw-rwx  2 root root  4096 Oct 24 15:14 scripts\r\n",
      "-rwxrw-rwx  1 root root   810 Oct 24 15:14 setup.py\r\n",
      "drwxrw-rwx  4 root root  4096 Oct 24 15:14 test\r\n"
     ]
    }
   ],
   "source": [
    "! ls -la ../../joeynmt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "! sudo chmod 777 ../../joeynmt/models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6ZBPFwT94WpI"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:541: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:542: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:543: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:544: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:545: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:550: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n",
      "2020-02-20 12:03:17,437 Hello! This is Joey-NMT.\n",
      "2020-02-20 12:03:17,444 Total params: 12144896\n",
      "2020-02-20 12:03:17,445 Trainable parameters: ['decoder.layer_norm.bias', 'decoder.layer_norm.weight', 'decoder.layers.0.dec_layer_norm.bias', 'decoder.layers.0.dec_layer_norm.weight', 'decoder.layers.0.feed_forward.layer_norm.bias', 'decoder.layers.0.feed_forward.layer_norm.weight', 'decoder.layers.0.feed_forward.pwff_layer.0.bias', 'decoder.layers.0.feed_forward.pwff_layer.0.weight', 'decoder.layers.0.feed_forward.pwff_layer.3.bias', 'decoder.layers.0.feed_forward.pwff_layer.3.weight', 'decoder.layers.0.src_trg_att.k_layer.bias', 'decoder.layers.0.src_trg_att.k_layer.weight', 'decoder.layers.0.src_trg_att.output_layer.bias', 'decoder.layers.0.src_trg_att.output_layer.weight', 'decoder.layers.0.src_trg_att.q_layer.bias', 'decoder.layers.0.src_trg_att.q_layer.weight', 'decoder.layers.0.src_trg_att.v_layer.bias', 'decoder.layers.0.src_trg_att.v_layer.weight', 'decoder.layers.0.trg_trg_att.k_layer.bias', 'decoder.layers.0.trg_trg_att.k_layer.weight', 'decoder.layers.0.trg_trg_att.output_layer.bias', 'decoder.layers.0.trg_trg_att.output_layer.weight', 'decoder.layers.0.trg_trg_att.q_layer.bias', 'decoder.layers.0.trg_trg_att.q_layer.weight', 'decoder.layers.0.trg_trg_att.v_layer.bias', 'decoder.layers.0.trg_trg_att.v_layer.weight', 'decoder.layers.0.x_layer_norm.bias', 'decoder.layers.0.x_layer_norm.weight', 'decoder.layers.1.dec_layer_norm.bias', 'decoder.layers.1.dec_layer_norm.weight', 'decoder.layers.1.feed_forward.layer_norm.bias', 'decoder.layers.1.feed_forward.layer_norm.weight', 'decoder.layers.1.feed_forward.pwff_layer.0.bias', 'decoder.layers.1.feed_forward.pwff_layer.0.weight', 'decoder.layers.1.feed_forward.pwff_layer.3.bias', 'decoder.layers.1.feed_forward.pwff_layer.3.weight', 'decoder.layers.1.src_trg_att.k_layer.bias', 'decoder.layers.1.src_trg_att.k_layer.weight', 'decoder.layers.1.src_trg_att.output_layer.bias', 'decoder.layers.1.src_trg_att.output_layer.weight', 'decoder.layers.1.src_trg_att.q_layer.bias', 'decoder.layers.1.src_trg_att.q_layer.weight', 'decoder.layers.1.src_trg_att.v_layer.bias', 'decoder.layers.1.src_trg_att.v_layer.weight', 'decoder.layers.1.trg_trg_att.k_layer.bias', 'decoder.layers.1.trg_trg_att.k_layer.weight', 'decoder.layers.1.trg_trg_att.output_layer.bias', 'decoder.layers.1.trg_trg_att.output_layer.weight', 'decoder.layers.1.trg_trg_att.q_layer.bias', 'decoder.layers.1.trg_trg_att.q_layer.weight', 'decoder.layers.1.trg_trg_att.v_layer.bias', 'decoder.layers.1.trg_trg_att.v_layer.weight', 'decoder.layers.1.x_layer_norm.bias', 'decoder.layers.1.x_layer_norm.weight', 'decoder.layers.2.dec_layer_norm.bias', 'decoder.layers.2.dec_layer_norm.weight', 'decoder.layers.2.feed_forward.layer_norm.bias', 'decoder.layers.2.feed_forward.layer_norm.weight', 'decoder.layers.2.feed_forward.pwff_layer.0.bias', 'decoder.layers.2.feed_forward.pwff_layer.0.weight', 'decoder.layers.2.feed_forward.pwff_layer.3.bias', 'decoder.layers.2.feed_forward.pwff_layer.3.weight', 'decoder.layers.2.src_trg_att.k_layer.bias', 'decoder.layers.2.src_trg_att.k_layer.weight', 'decoder.layers.2.src_trg_att.output_layer.bias', 'decoder.layers.2.src_trg_att.output_layer.weight', 'decoder.layers.2.src_trg_att.q_layer.bias', 'decoder.layers.2.src_trg_att.q_layer.weight', 'decoder.layers.2.src_trg_att.v_layer.bias', 'decoder.layers.2.src_trg_att.v_layer.weight', 'decoder.layers.2.trg_trg_att.k_layer.bias', 'decoder.layers.2.trg_trg_att.k_layer.weight', 'decoder.layers.2.trg_trg_att.output_layer.bias', 'decoder.layers.2.trg_trg_att.output_layer.weight', 'decoder.layers.2.trg_trg_att.q_layer.bias', 'decoder.layers.2.trg_trg_att.q_layer.weight', 'decoder.layers.2.trg_trg_att.v_layer.bias', 'decoder.layers.2.trg_trg_att.v_layer.weight', 'decoder.layers.2.x_layer_norm.bias', 'decoder.layers.2.x_layer_norm.weight', 'decoder.layers.3.dec_layer_norm.bias', 'decoder.layers.3.dec_layer_norm.weight', 'decoder.layers.3.feed_forward.layer_norm.bias', 'decoder.layers.3.feed_forward.layer_norm.weight', 'decoder.layers.3.feed_forward.pwff_layer.0.bias', 'decoder.layers.3.feed_forward.pwff_layer.0.weight', 'decoder.layers.3.feed_forward.pwff_layer.3.bias', 'decoder.layers.3.feed_forward.pwff_layer.3.weight', 'decoder.layers.3.src_trg_att.k_layer.bias', 'decoder.layers.3.src_trg_att.k_layer.weight', 'decoder.layers.3.src_trg_att.output_layer.bias', 'decoder.layers.3.src_trg_att.output_layer.weight', 'decoder.layers.3.src_trg_att.q_layer.bias', 'decoder.layers.3.src_trg_att.q_layer.weight', 'decoder.layers.3.src_trg_att.v_layer.bias', 'decoder.layers.3.src_trg_att.v_layer.weight', 'decoder.layers.3.trg_trg_att.k_layer.bias', 'decoder.layers.3.trg_trg_att.k_layer.weight', 'decoder.layers.3.trg_trg_att.output_layer.bias', 'decoder.layers.3.trg_trg_att.output_layer.weight', 'decoder.layers.3.trg_trg_att.q_layer.bias', 'decoder.layers.3.trg_trg_att.q_layer.weight', 'decoder.layers.3.trg_trg_att.v_layer.bias', 'decoder.layers.3.trg_trg_att.v_layer.weight', 'decoder.layers.3.x_layer_norm.bias', 'decoder.layers.3.x_layer_norm.weight', 'decoder.layers.4.dec_layer_norm.bias', 'decoder.layers.4.dec_layer_norm.weight', 'decoder.layers.4.feed_forward.layer_norm.bias', 'decoder.layers.4.feed_forward.layer_norm.weight', 'decoder.layers.4.feed_forward.pwff_layer.0.bias', 'decoder.layers.4.feed_forward.pwff_layer.0.weight', 'decoder.layers.4.feed_forward.pwff_layer.3.bias', 'decoder.layers.4.feed_forward.pwff_layer.3.weight', 'decoder.layers.4.src_trg_att.k_layer.bias', 'decoder.layers.4.src_trg_att.k_layer.weight', 'decoder.layers.4.src_trg_att.output_layer.bias', 'decoder.layers.4.src_trg_att.output_layer.weight', 'decoder.layers.4.src_trg_att.q_layer.bias', 'decoder.layers.4.src_trg_att.q_layer.weight', 'decoder.layers.4.src_trg_att.v_layer.bias', 'decoder.layers.4.src_trg_att.v_layer.weight', 'decoder.layers.4.trg_trg_att.k_layer.bias', 'decoder.layers.4.trg_trg_att.k_layer.weight', 'decoder.layers.4.trg_trg_att.output_layer.bias', 'decoder.layers.4.trg_trg_att.output_layer.weight', 'decoder.layers.4.trg_trg_att.q_layer.bias', 'decoder.layers.4.trg_trg_att.q_layer.weight', 'decoder.layers.4.trg_trg_att.v_layer.bias', 'decoder.layers.4.trg_trg_att.v_layer.weight', 'decoder.layers.4.x_layer_norm.bias', 'decoder.layers.4.x_layer_norm.weight', 'decoder.layers.5.dec_layer_norm.bias', 'decoder.layers.5.dec_layer_norm.weight', 'decoder.layers.5.feed_forward.layer_norm.bias', 'decoder.layers.5.feed_forward.layer_norm.weight', 'decoder.layers.5.feed_forward.pwff_layer.0.bias', 'decoder.layers.5.feed_forward.pwff_layer.0.weight', 'decoder.layers.5.feed_forward.pwff_layer.3.bias', 'decoder.layers.5.feed_forward.pwff_layer.3.weight', 'decoder.layers.5.src_trg_att.k_layer.bias', 'decoder.layers.5.src_trg_att.k_layer.weight', 'decoder.layers.5.src_trg_att.output_layer.bias', 'decoder.layers.5.src_trg_att.output_layer.weight', 'decoder.layers.5.src_trg_att.q_layer.bias', 'decoder.layers.5.src_trg_att.q_layer.weight', 'decoder.layers.5.src_trg_att.v_layer.bias', 'decoder.layers.5.src_trg_att.v_layer.weight', 'decoder.layers.5.trg_trg_att.k_layer.bias', 'decoder.layers.5.trg_trg_att.k_layer.weight', 'decoder.layers.5.trg_trg_att.output_layer.bias', 'decoder.layers.5.trg_trg_att.output_layer.weight', 'decoder.layers.5.trg_trg_att.q_layer.bias', 'decoder.layers.5.trg_trg_att.q_layer.weight', 'decoder.layers.5.trg_trg_att.v_layer.bias', 'decoder.layers.5.trg_trg_att.v_layer.weight', 'decoder.layers.5.x_layer_norm.bias', 'decoder.layers.5.x_layer_norm.weight', 'encoder.layer_norm.bias', 'encoder.layer_norm.weight', 'encoder.layers.0.feed_forward.layer_norm.bias', 'encoder.layers.0.feed_forward.layer_norm.weight', 'encoder.layers.0.feed_forward.pwff_layer.0.bias', 'encoder.layers.0.feed_forward.pwff_layer.0.weight', 'encoder.layers.0.feed_forward.pwff_layer.3.bias', 'encoder.layers.0.feed_forward.pwff_layer.3.weight', 'encoder.layers.0.layer_norm.bias', 'encoder.layers.0.layer_norm.weight', 'encoder.layers.0.src_src_att.k_layer.bias', 'encoder.layers.0.src_src_att.k_layer.weight', 'encoder.layers.0.src_src_att.output_layer.bias', 'encoder.layers.0.src_src_att.output_layer.weight', 'encoder.layers.0.src_src_att.q_layer.bias', 'encoder.layers.0.src_src_att.q_layer.weight', 'encoder.layers.0.src_src_att.v_layer.bias', 'encoder.layers.0.src_src_att.v_layer.weight', 'encoder.layers.1.feed_forward.layer_norm.bias', 'encoder.layers.1.feed_forward.layer_norm.weight', 'encoder.layers.1.feed_forward.pwff_layer.0.bias', 'encoder.layers.1.feed_forward.pwff_layer.0.weight', 'encoder.layers.1.feed_forward.pwff_layer.3.bias', 'encoder.layers.1.feed_forward.pwff_layer.3.weight', 'encoder.layers.1.layer_norm.bias', 'encoder.layers.1.layer_norm.weight', 'encoder.layers.1.src_src_att.k_layer.bias', 'encoder.layers.1.src_src_att.k_layer.weight', 'encoder.layers.1.src_src_att.output_layer.bias', 'encoder.layers.1.src_src_att.output_layer.weight', 'encoder.layers.1.src_src_att.q_layer.bias', 'encoder.layers.1.src_src_att.q_layer.weight', 'encoder.layers.1.src_src_att.v_layer.bias', 'encoder.layers.1.src_src_att.v_layer.weight', 'encoder.layers.2.feed_forward.layer_norm.bias', 'encoder.layers.2.feed_forward.layer_norm.weight', 'encoder.layers.2.feed_forward.pwff_layer.0.bias', 'encoder.layers.2.feed_forward.pwff_layer.0.weight', 'encoder.layers.2.feed_forward.pwff_layer.3.bias', 'encoder.layers.2.feed_forward.pwff_layer.3.weight', 'encoder.layers.2.layer_norm.bias', 'encoder.layers.2.layer_norm.weight', 'encoder.layers.2.src_src_att.k_layer.bias', 'encoder.layers.2.src_src_att.k_layer.weight', 'encoder.layers.2.src_src_att.output_layer.bias', 'encoder.layers.2.src_src_att.output_layer.weight', 'encoder.layers.2.src_src_att.q_layer.bias', 'encoder.layers.2.src_src_att.q_layer.weight', 'encoder.layers.2.src_src_att.v_layer.bias', 'encoder.layers.2.src_src_att.v_layer.weight', 'encoder.layers.3.feed_forward.layer_norm.bias', 'encoder.layers.3.feed_forward.layer_norm.weight', 'encoder.layers.3.feed_forward.pwff_layer.0.bias', 'encoder.layers.3.feed_forward.pwff_layer.0.weight', 'encoder.layers.3.feed_forward.pwff_layer.3.bias', 'encoder.layers.3.feed_forward.pwff_layer.3.weight', 'encoder.layers.3.layer_norm.bias', 'encoder.layers.3.layer_norm.weight', 'encoder.layers.3.src_src_att.k_layer.bias', 'encoder.layers.3.src_src_att.k_layer.weight', 'encoder.layers.3.src_src_att.output_layer.bias', 'encoder.layers.3.src_src_att.output_layer.weight', 'encoder.layers.3.src_src_att.q_layer.bias', 'encoder.layers.3.src_src_att.q_layer.weight', 'encoder.layers.3.src_src_att.v_layer.bias', 'encoder.layers.3.src_src_att.v_layer.weight', 'encoder.layers.4.feed_forward.layer_norm.bias', 'encoder.layers.4.feed_forward.layer_norm.weight', 'encoder.layers.4.feed_forward.pwff_layer.0.bias', 'encoder.layers.4.feed_forward.pwff_layer.0.weight', 'encoder.layers.4.feed_forward.pwff_layer.3.bias', 'encoder.layers.4.feed_forward.pwff_layer.3.weight', 'encoder.layers.4.layer_norm.bias', 'encoder.layers.4.layer_norm.weight', 'encoder.layers.4.src_src_att.k_layer.bias', 'encoder.layers.4.src_src_att.k_layer.weight', 'encoder.layers.4.src_src_att.output_layer.bias', 'encoder.layers.4.src_src_att.output_layer.weight', 'encoder.layers.4.src_src_att.q_layer.bias', 'encoder.layers.4.src_src_att.q_layer.weight', 'encoder.layers.4.src_src_att.v_layer.bias', 'encoder.layers.4.src_src_att.v_layer.weight', 'encoder.layers.5.feed_forward.layer_norm.bias', 'encoder.layers.5.feed_forward.layer_norm.weight', 'encoder.layers.5.feed_forward.pwff_layer.0.bias', 'encoder.layers.5.feed_forward.pwff_layer.0.weight', 'encoder.layers.5.feed_forward.pwff_layer.3.bias', 'encoder.layers.5.feed_forward.pwff_layer.3.weight', 'encoder.layers.5.layer_norm.bias', 'encoder.layers.5.layer_norm.weight', 'encoder.layers.5.src_src_att.k_layer.bias', 'encoder.layers.5.src_src_att.k_layer.weight', 'encoder.layers.5.src_src_att.output_layer.bias', 'encoder.layers.5.src_src_att.output_layer.weight', 'encoder.layers.5.src_src_att.q_layer.bias', 'encoder.layers.5.src_src_att.q_layer.weight', 'encoder.layers.5.src_src_att.v_layer.bias', 'encoder.layers.5.src_src_att.v_layer.weight', 'src_embed.lut.weight']\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 12:03:20,118 cfg.name                           : enluo_transformer\n",
      "2020-02-20 12:03:20,118 cfg.data.src                       : en\n",
      "2020-02-20 12:03:20,118 cfg.data.trg                       : luo\n",
      "2020-02-20 12:03:20,118 cfg.data.train                     : data/enluo/train.bpe\n",
      "2020-02-20 12:03:20,118 cfg.data.dev                       : data/enluo/dev.bpe\n",
      "2020-02-20 12:03:20,118 cfg.data.test                      : data/enluo/test.bpe\n",
      "2020-02-20 12:03:20,118 cfg.data.level                     : bpe\n",
      "2020-02-20 12:03:20,118 cfg.data.lowercase                 : False\n",
      "2020-02-20 12:03:20,118 cfg.data.max_sent_length           : 100\n",
      "2020-02-20 12:03:20,118 cfg.data.src_vocab                 : data/enluo/vocab.txt\n",
      "2020-02-20 12:03:20,118 cfg.data.trg_vocab                 : data/enluo/vocab.txt\n",
      "2020-02-20 12:03:20,118 cfg.testing.beam_size              : 5\n",
      "2020-02-20 12:03:20,118 cfg.testing.alpha                  : 1.0\n",
      "2020-02-20 12:03:20,119 cfg.training.random_seed           : 42\n",
      "2020-02-20 12:03:20,119 cfg.training.optimizer             : adam\n",
      "2020-02-20 12:03:20,119 cfg.training.normalization         : tokens\n",
      "2020-02-20 12:03:20,119 cfg.training.adam_betas            : [0.9, 0.999]\n",
      "2020-02-20 12:03:20,119 cfg.training.scheduling            : plateau\n",
      "2020-02-20 12:03:20,119 cfg.training.patience              : 5\n",
      "2020-02-20 12:03:20,119 cfg.training.learning_rate_factor  : 0.5\n",
      "2020-02-20 12:03:20,119 cfg.training.learning_rate_warmup  : 1000\n",
      "2020-02-20 12:03:20,119 cfg.training.decrease_factor       : 0.7\n",
      "2020-02-20 12:03:20,119 cfg.training.loss                  : crossentropy\n",
      "2020-02-20 12:03:20,119 cfg.training.learning_rate         : 0.0003\n",
      "2020-02-20 12:03:20,119 cfg.training.learning_rate_min     : 1e-08\n",
      "2020-02-20 12:03:20,119 cfg.training.weight_decay          : 0.0\n",
      "2020-02-20 12:03:20,119 cfg.training.label_smoothing       : 0.1\n",
      "2020-02-20 12:03:20,119 cfg.training.batch_size            : 4096\n",
      "2020-02-20 12:03:20,119 cfg.training.batch_type            : token\n",
      "2020-02-20 12:03:20,120 cfg.training.eval_batch_size       : 3600\n",
      "2020-02-20 12:03:20,120 cfg.training.eval_batch_type       : token\n",
      "2020-02-20 12:03:20,120 cfg.training.batch_multiplier      : 1\n",
      "2020-02-20 12:03:20,120 cfg.training.early_stopping_metric : ppl\n",
      "2020-02-20 12:03:20,120 cfg.training.epochs                : 30\n",
      "2020-02-20 12:03:20,120 cfg.training.validation_freq       : 1000\n",
      "2020-02-20 12:03:20,120 cfg.training.logging_freq          : 100\n",
      "2020-02-20 12:03:20,120 cfg.training.eval_metric           : bleu\n",
      "2020-02-20 12:03:20,120 cfg.training.model_dir             : models/enluo_transformer\n",
      "2020-02-20 12:03:20,120 cfg.training.overwrite             : False\n",
      "2020-02-20 12:03:20,120 cfg.training.shuffle               : True\n",
      "2020-02-20 12:03:20,120 cfg.training.use_cuda              : True\n",
      "2020-02-20 12:03:20,120 cfg.training.max_output_length     : 100\n",
      "2020-02-20 12:03:20,120 cfg.training.print_valid_sents     : [0, 1, 2, 3]\n",
      "2020-02-20 12:03:20,120 cfg.training.keep_last_ckpts       : 3\n",
      "2020-02-20 12:03:20,121 cfg.model.initializer              : xavier\n",
      "2020-02-20 12:03:20,121 cfg.model.bias_initializer         : zeros\n",
      "2020-02-20 12:03:20,121 cfg.model.init_gain                : 1.0\n",
      "2020-02-20 12:03:20,121 cfg.model.embed_initializer        : xavier\n",
      "2020-02-20 12:03:20,121 cfg.model.embed_init_gain          : 1.0\n",
      "2020-02-20 12:03:20,121 cfg.model.tied_embeddings          : True\n",
      "2020-02-20 12:03:20,121 cfg.model.tied_softmax             : True\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.type             : transformer\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.num_layers       : 6\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.num_heads        : 4\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.embeddings.embedding_dim : 256\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.embeddings.scale : True\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.embeddings.dropout : 0.2\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.hidden_size      : 256\n",
      "2020-02-20 12:03:20,121 cfg.model.encoder.ff_size          : 1024\n",
      "2020-02-20 12:03:20,122 cfg.model.encoder.dropout          : 0.3\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.type             : transformer\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.num_layers       : 6\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.num_heads        : 4\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.embeddings.embedding_dim : 256\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.embeddings.scale : True\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.embeddings.dropout : 0.2\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.hidden_size      : 256\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.ff_size          : 1024\n",
      "2020-02-20 12:03:20,122 cfg.model.decoder.dropout          : 0.3\n",
      "2020-02-20 12:03:20,122 Data set sizes: \n",
      "\ttrain 136375,\n",
      "\tvalid 1000,\n",
      "\ttest 2705\n",
      "2020-02-20 12:03:20,122 First training example:\n",
      "\t[SRC] Paul wrote : “ Let the su@@ n not set with you in a pro@@ v@@ ok@@ ed st@@ ate . ”\n",
      "\t[TRG] Paulo nondiko kama : “ Kik uwe chieng ’ p@@ odhi ka pod un gi mirima . ”\n",
      "2020-02-20 12:03:20,122 First 10 words (src): (0) <unk> (1) <pad> (2) <s> (3) </s> (4) . (5) , (6) the (7) to (8) e (9) ne\n",
      "2020-02-20 12:03:20,123 First 10 words (trg): (0) <unk> (1) <pad> (2) <s> (3) </s> (4) . (5) , (6) the (7) to (8) e (9) ne\n",
      "2020-02-20 12:03:20,123 Number of Src words (types): 4237\n",
      "2020-02-20 12:03:20,123 Number of Trg words (types): 4237\n",
      "2020-02-20 12:03:20,123 Model(\n",
      "\tencoder=TransformerEncoder(num_layers=6, num_heads=4),\n",
      "\tdecoder=TransformerDecoder(num_layers=6, num_heads=4),\n",
      "\tsrc_embed=Embeddings(embedding_dim=256, vocab_size=4237),\n",
      "\ttrg_embed=Embeddings(embedding_dim=256, vocab_size=4237))\n",
      "2020-02-20 12:03:20,127 EPOCH 1\n",
      "2020-02-20 12:03:49,513 Epoch   1 Step:      100 Batch Loss:     5.409370 Tokens per Sec:     7544, Lr: 0.000300\n",
      "2020-02-20 12:04:18,216 Epoch   1 Step:      200 Batch Loss:     5.257635 Tokens per Sec:     7619, Lr: 0.000300\n",
      "2020-02-20 12:04:47,204 Epoch   1 Step:      300 Batch Loss:     5.064311 Tokens per Sec:     7625, Lr: 0.000300\n",
      "2020-02-20 12:05:15,922 Epoch   1 Step:      400 Batch Loss:     4.967086 Tokens per Sec:     7592, Lr: 0.000300\n",
      "2020-02-20 12:05:45,060 Epoch   1 Step:      500 Batch Loss:     4.752475 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 12:06:13,895 Epoch   1 Step:      600 Batch Loss:     4.454239 Tokens per Sec:     7485, Lr: 0.000300\n",
      "2020-02-20 12:06:42,961 Epoch   1 Step:      700 Batch Loss:     4.371806 Tokens per Sec:     7564, Lr: 0.000300\n",
      "2020-02-20 12:07:11,959 Epoch   1 Step:      800 Batch Loss:     4.450417 Tokens per Sec:     7661, Lr: 0.000300\n",
      "2020-02-20 12:07:40,856 Epoch   1 Step:      900 Batch Loss:     4.141620 Tokens per Sec:     7539, Lr: 0.000300\n",
      "2020-02-20 12:08:09,675 Epoch   1 Step:     1000 Batch Loss:     4.208232 Tokens per Sec:     7618, Lr: 0.000300\n",
      "2020-02-20 12:09:37,993 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:09:37,993 Saving new checkpoint.\n",
      "2020-02-20 12:09:38,204 Example #0\n",
      "2020-02-20 12:09:38,204 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:09:38,204 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:09:38,204 \tHypothesis: Mano kaka wanyalo bedo gi Jehova , Jehova nyalo bedo gi Jehova .\n",
      "2020-02-20 12:09:38,204 Example #1\n",
      "2020-02-20 12:09:38,205 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:09:38,205 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:09:38,205 \tHypothesis: Kata kamano , ne ok wabedo gi gik ma ne ok ok wabedo gi Nyasaye . ”\n",
      "2020-02-20 12:09:38,205 Example #2\n",
      "2020-02-20 12:09:38,205 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:09:38,205 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:09:38,205 \tHypothesis: Kata kamano , wanyalo bedo gi gik ma ne wabedo gi gik ma ne watimo kamano .\n",
      "2020-02-20 12:09:38,205 Example #3\n",
      "2020-02-20 12:09:38,205 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:09:38,205 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:09:38,205 \tHypothesis: Ne en ni “ Nyasaye nowacho ni “ Nyasaye , ” ma ne en gima ne en gima ne en gima ne en gima ne en gima ne en gima ne en gima ne en gima ne en gima ne en . ”\n",
      "2020-02-20 12:09:38,206 Validation result at epoch   1, step     1000: bleu:   1.01, loss: 98175.9766, ppl:  53.6843, duration: 88.5302s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 12:10:07,191 Epoch   1 Step:     1100 Batch Loss:     3.762488 Tokens per Sec:     7596, Lr: 0.000300\n",
      "2020-02-20 12:10:36,062 Epoch   1 Step:     1200 Batch Loss:     4.396629 Tokens per Sec:     7597, Lr: 0.000300\n",
      "2020-02-20 12:11:04,709 Epoch   1 Step:     1300 Batch Loss:     4.256168 Tokens per Sec:     7514, Lr: 0.000300\n",
      "2020-02-20 12:11:33,621 Epoch   1 Step:     1400 Batch Loss:     3.945779 Tokens per Sec:     7441, Lr: 0.000300\n",
      "2020-02-20 12:12:02,358 Epoch   1 Step:     1500 Batch Loss:     3.935235 Tokens per Sec:     7623, Lr: 0.000300\n",
      "2020-02-20 12:12:07,028 Epoch   1: total training loss 6731.63\n",
      "2020-02-20 12:12:07,029 EPOCH 2\n",
      "2020-02-20 12:12:31,139 Epoch   2 Step:     1600 Batch Loss:     3.035681 Tokens per Sec:     7441, Lr: 0.000300\n",
      "2020-02-20 12:12:59,846 Epoch   2 Step:     1700 Batch Loss:     3.587291 Tokens per Sec:     7593, Lr: 0.000300\n",
      "2020-02-20 12:13:28,871 Epoch   2 Step:     1800 Batch Loss:     3.648604 Tokens per Sec:     7519, Lr: 0.000300\n",
      "2020-02-20 12:13:57,438 Epoch   2 Step:     1900 Batch Loss:     2.682493 Tokens per Sec:     7478, Lr: 0.000300\n",
      "2020-02-20 12:14:26,622 Epoch   2 Step:     2000 Batch Loss:     3.596354 Tokens per Sec:     7753, Lr: 0.000300\n",
      "2020-02-20 12:15:54,913 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:15:54,913 Saving new checkpoint.\n",
      "2020-02-20 12:15:55,109 Example #0\n",
      "2020-02-20 12:15:55,110 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:15:55,110 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:15:55,110 \tHypothesis: Obiro konyowa mondo wakaw okang ’ ma Jehova dwaro ni wayie motegno e wi Nyasaye .\n",
      "2020-02-20 12:15:55,110 Example #1\n",
      "2020-02-20 12:15:55,110 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:15:55,110 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:15:55,110 \tHypothesis: Mano kaka ne ok en gima ne ok nyal bedo gi yie kendo ne ok giyie gi Nyasaye . ”\n",
      "2020-02-20 12:15:55,110 Example #2\n",
      "2020-02-20 12:15:55,110 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:15:55,110 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:15:55,111 \tHypothesis: Kokalo kuom weche ma Muma nyalo konyowa mondo wakaw okang ’ mar Muma , kendo waparo ni wayie motegno e wi gik ma nyalo miyo ng’ato obed gi geno mar timo .\n",
      "2020-02-20 12:15:55,111 Example #3\n",
      "2020-02-20 12:15:55,111 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:15:55,111 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:15:55,111 \tHypothesis: Ne giwuoyo kuom “ oganda Nyasaye ” ma ne dhi timore e wi Yesu , kendo ne dhi bedo gi geno mar Yesu .\n",
      "2020-02-20 12:15:55,111 Validation result at epoch   2, step     2000: bleu:   2.74, loss: 83177.7891, ppl:  29.2134, duration: 88.4880s\n",
      "2020-02-20 12:16:23,645 Epoch   2 Step:     2100 Batch Loss:     3.629211 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 12:16:52,681 Epoch   2 Step:     2200 Batch Loss:     3.622419 Tokens per Sec:     7664, Lr: 0.000300\n",
      "2020-02-20 12:17:21,336 Epoch   2 Step:     2300 Batch Loss:     3.560351 Tokens per Sec:     7579, Lr: 0.000300\n",
      "2020-02-20 12:17:50,095 Epoch   2 Step:     2400 Batch Loss:     3.292273 Tokens per Sec:     7494, Lr: 0.000300\n",
      "2020-02-20 12:18:19,258 Epoch   2 Step:     2500 Batch Loss:     3.300863 Tokens per Sec:     7592, Lr: 0.000300\n",
      "2020-02-20 12:18:48,170 Epoch   2 Step:     2600 Batch Loss:     2.710793 Tokens per Sec:     7468, Lr: 0.000300\n",
      "2020-02-20 12:19:16,980 Epoch   2 Step:     2700 Batch Loss:     3.404884 Tokens per Sec:     7585, Lr: 0.000300\n",
      "2020-02-20 12:19:45,680 Epoch   2 Step:     2800 Batch Loss:     2.982437 Tokens per Sec:     7570, Lr: 0.000300\n",
      "2020-02-20 12:20:14,533 Epoch   2 Step:     2900 Batch Loss:     2.992396 Tokens per Sec:     7690, Lr: 0.000300\n",
      "2020-02-20 12:20:43,253 Epoch   2 Step:     3000 Batch Loss:     2.899553 Tokens per Sec:     7606, Lr: 0.000300\n",
      "2020-02-20 12:22:11,433 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:22:11,433 Saving new checkpoint.\n",
      "2020-02-20 12:22:11,627 Example #0\n",
      "2020-02-20 12:22:11,627 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:22:11,627 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:22:11,627 \tHypothesis: Jonyuol ma Jehova dwaro ni wabed gi yie motegno e wi Nyasaye , onego omi wabed gi paro ma Nyasaye nigo .\n",
      "2020-02-20 12:22:11,627 Example #1\n",
      "2020-02-20 12:22:11,627 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:22:11,627 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:22:11,628 \tHypothesis: Bende , ne awinjo ni ne amor gi Nyasaye kendo ne amor . ”\n",
      "2020-02-20 12:22:11,628 Example #2\n",
      "2020-02-20 12:22:11,628 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:22:11,628 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:22:11,628 \tHypothesis: Tem ane kaka wanyalo ng’eyo weche manie Muma e wi gik ma nyalo miyo ng’ato obed gi chuny mar timo .\n",
      "2020-02-20 12:22:11,628 Example #3\n",
      "2020-02-20 12:22:11,628 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:22:11,628 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:22:11,628 \tHypothesis: Ne giparo ni “ Nyasaye non ji duto ma ne ochiwo e wi roho maler . ”\n",
      "2020-02-20 12:22:11,628 Validation result at epoch   2, step     3000: bleu:   4.86, loss: 75435.2109, ppl:  21.3382, duration: 88.3752s\n",
      "2020-02-20 12:22:22,268 Epoch   2: total training loss 5145.10\n",
      "2020-02-20 12:22:22,269 EPOCH 3\n",
      "2020-02-20 12:22:40,557 Epoch   3 Step:     3100 Batch Loss:     3.012194 Tokens per Sec:     7509, Lr: 0.000300\n",
      "2020-02-20 12:23:09,222 Epoch   3 Step:     3200 Batch Loss:     2.736670 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 12:23:38,282 Epoch   3 Step:     3300 Batch Loss:     3.120730 Tokens per Sec:     7632, Lr: 0.000300\n",
      "2020-02-20 12:24:06,960 Epoch   3 Step:     3400 Batch Loss:     3.273985 Tokens per Sec:     7542, Lr: 0.000300\n",
      "2020-02-20 12:24:35,743 Epoch   3 Step:     3500 Batch Loss:     3.094509 Tokens per Sec:     7650, Lr: 0.000300\n",
      "2020-02-20 12:25:04,392 Epoch   3 Step:     3600 Batch Loss:     3.046881 Tokens per Sec:     7579, Lr: 0.000300\n",
      "2020-02-20 12:25:33,262 Epoch   3 Step:     3700 Batch Loss:     3.162375 Tokens per Sec:     7671, Lr: 0.000300\n",
      "2020-02-20 12:26:02,164 Epoch   3 Step:     3800 Batch Loss:     3.190458 Tokens per Sec:     7655, Lr: 0.000300\n",
      "2020-02-20 12:26:30,733 Epoch   3 Step:     3900 Batch Loss:     3.196185 Tokens per Sec:     7489, Lr: 0.000300\n",
      "2020-02-20 12:26:59,709 Epoch   3 Step:     4000 Batch Loss:     3.097071 Tokens per Sec:     7623, Lr: 0.000300\n",
      "2020-02-20 12:28:27,888 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:28:27,889 Saving new checkpoint.\n",
      "2020-02-20 12:28:28,107 Example #0\n",
      "2020-02-20 12:28:28,107 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:28:28,108 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:28:28,108 \tHypothesis: Ne giyie mondo Jehova otagi mondo omi giwinj tiend wach Nyasaye .\n",
      "2020-02-20 12:28:28,108 Example #1\n",
      "2020-02-20 12:28:28,108 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:28:28,108 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:28:28,108 \tHypothesis: Bende , ne awachona ni , kendo ne awinjo ka an gi yie . ”\n",
      "2020-02-20 12:28:28,108 Example #2\n",
      "2020-02-20 12:28:28,108 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:28:28,108 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:28:28,108 \tHypothesis: Tem ane gik ma Muma wacho e wi gik ma nyalo timore e wi gik ma nyalo miyo ng’ato obed gi chuny maler .\n",
      "2020-02-20 12:28:28,109 Example #3\n",
      "2020-02-20 12:28:28,109 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:28:28,109 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:28:28,109 \tHypothesis: Ne owachone ni “ Nyasaye duto ma ne osetiedho ji duto ma ne ni e ndalo mag giko . ”\n",
      "2020-02-20 12:28:28,109 Validation result at epoch   3, step     4000: bleu:   7.70, loss: 69503.4297, ppl:  16.7742, duration: 88.3992s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 12:28:56,806 Epoch   3 Step:     4100 Batch Loss:     2.952160 Tokens per Sec:     7502, Lr: 0.000300\n",
      "2020-02-20 12:29:25,371 Epoch   3 Step:     4200 Batch Loss:     3.406082 Tokens per Sec:     7499, Lr: 0.000300\n",
      "2020-02-20 12:29:54,297 Epoch   3 Step:     4300 Batch Loss:     3.000547 Tokens per Sec:     7585, Lr: 0.000300\n",
      "2020-02-20 12:30:22,684 Epoch   3 Step:     4400 Batch Loss:     3.101769 Tokens per Sec:     7585, Lr: 0.000300\n",
      "2020-02-20 12:30:51,661 Epoch   3 Step:     4500 Batch Loss:     2.918369 Tokens per Sec:     7768, Lr: 0.000300\n",
      "2020-02-20 12:31:07,842 Epoch   3: total training loss 4516.30\n",
      "2020-02-20 12:31:07,843 EPOCH 4\n",
      "2020-02-20 12:31:20,580 Epoch   4 Step:     4600 Batch Loss:     2.786486 Tokens per Sec:     7292, Lr: 0.000300\n",
      "2020-02-20 12:31:49,217 Epoch   4 Step:     4700 Batch Loss:     2.668926 Tokens per Sec:     7503, Lr: 0.000300\n",
      "2020-02-20 12:32:18,079 Epoch   4 Step:     4800 Batch Loss:     2.563730 Tokens per Sec:     7651, Lr: 0.000300\n",
      "2020-02-20 12:32:47,159 Epoch   4 Step:     4900 Batch Loss:     2.711700 Tokens per Sec:     7726, Lr: 0.000300\n",
      "2020-02-20 12:33:16,085 Epoch   4 Step:     5000 Batch Loss:     2.843105 Tokens per Sec:     7747, Lr: 0.000300\n",
      "2020-02-20 12:34:44,283 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:34:44,284 Saving new checkpoint.\n",
      "2020-02-20 12:34:44,499 Example #0\n",
      "2020-02-20 12:34:44,500 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:34:44,500 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:34:44,500 \tHypothesis: Ne gitemo nyiso ni Jehova dwaro ni giher ni giher ni giluw chike Nyasaye .\n",
      "2020-02-20 12:34:44,500 Example #1\n",
      "2020-02-20 12:34:44,500 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:34:44,500 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:34:44,500 \tHypothesis: Bende , amor ahinya mondo awinj luor kendo amor gi osiepena . ”\n",
      "2020-02-20 12:34:44,500 Example #2\n",
      "2020-02-20 12:34:44,501 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:34:44,501 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:34:44,501 \tHypothesis: Bedo gi paro matut kuom weche manie Ndiko nyalo miyo waket pachwa kuom gik ma nyalo miyo ng’ato obed gi chuny maler .\n",
      "2020-02-20 12:34:44,501 Example #3\n",
      "2020-02-20 12:34:44,501 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:34:44,501 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:34:44,501 \tHypothesis: Ne owuoyo kuom weche ma ne okor e wi “ gik moko duto ma Nyasaye notiyogo e kinde machon . ”\n",
      "2020-02-20 12:34:44,501 Validation result at epoch   4, step     5000: bleu:   9.56, loss: 65853.0234, ppl:  14.4651, duration: 88.4158s\n",
      "2020-02-20 12:35:12,990 Epoch   4 Step:     5100 Batch Loss:     2.572210 Tokens per Sec:     7391, Lr: 0.000300\n",
      "2020-02-20 12:35:42,102 Epoch   4 Step:     5200 Batch Loss:     2.858901 Tokens per Sec:     7599, Lr: 0.000300\n",
      "2020-02-20 12:36:11,060 Epoch   4 Step:     5300 Batch Loss:     2.993764 Tokens per Sec:     7639, Lr: 0.000300\n",
      "2020-02-20 12:36:39,920 Epoch   4 Step:     5400 Batch Loss:     2.127422 Tokens per Sec:     7607, Lr: 0.000300\n",
      "2020-02-20 12:37:08,685 Epoch   4 Step:     5500 Batch Loss:     2.690960 Tokens per Sec:     7572, Lr: 0.000300\n",
      "2020-02-20 12:37:37,664 Epoch   4 Step:     5600 Batch Loss:     3.039052 Tokens per Sec:     7527, Lr: 0.000300\n",
      "2020-02-20 12:38:06,762 Epoch   4 Step:     5700 Batch Loss:     2.761341 Tokens per Sec:     7523, Lr: 0.000300\n",
      "2020-02-20 12:38:35,581 Epoch   4 Step:     5800 Batch Loss:     2.505060 Tokens per Sec:     7476, Lr: 0.000300\n",
      "2020-02-20 12:39:04,462 Epoch   4 Step:     5900 Batch Loss:     2.251678 Tokens per Sec:     7545, Lr: 0.000300\n",
      "2020-02-20 12:39:33,184 Epoch   4 Step:     6000 Batch Loss:     2.194507 Tokens per Sec:     7611, Lr: 0.000300\n",
      "2020-02-20 12:41:01,319 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:41:01,320 Saving new checkpoint.\n",
      "2020-02-20 12:41:01,534 Example #0\n",
      "2020-02-20 12:41:01,534 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:41:01,534 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:41:01,534 \tHypothesis: Geno ma Jehova dwaro ni watim gik ma dwaro ni watim mondo omi dhano obed gi paro makare .\n",
      "2020-02-20 12:41:01,535 Example #1\n",
      "2020-02-20 12:41:01,535 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:41:01,535 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:41:01,535 \tHypothesis: Bende , amor ahinya mondo awinj tiend luor kod luor ma en - go mar Nyasaye . ”\n",
      "2020-02-20 12:41:01,535 Example #2\n",
      "2020-02-20 12:41:01,535 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:41:01,535 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:41:01,535 \tHypothesis: Kuom adier , ka waparo matut kuom Ndiko nyalo miyo ng’ato obed gi chuny maler .\n",
      "2020-02-20 12:41:01,535 Example #3\n",
      "2020-02-20 12:41:01,536 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:41:01,536 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:41:01,536 \tHypothesis: Ne ooro weche ma ne okor e wi “ gik moko duto ma Nyasaye nowacho e wi ndalo mag giko . ”\n",
      "2020-02-20 12:41:01,536 Validation result at epoch   4, step     6000: bleu:  11.00, loss: 63031.2188, ppl:  12.9004, duration: 88.3515s\n",
      "2020-02-20 12:41:23,076 Epoch   4: total training loss 4153.75\n",
      "2020-02-20 12:41:23,076 EPOCH 5\n",
      "2020-02-20 12:41:30,384 Epoch   5 Step:     6100 Batch Loss:     2.903340 Tokens per Sec:     7358, Lr: 0.000300\n",
      "2020-02-20 12:41:59,188 Epoch   5 Step:     6200 Batch Loss:     2.727892 Tokens per Sec:     7711, Lr: 0.000300\n",
      "2020-02-20 12:42:27,997 Epoch   5 Step:     6300 Batch Loss:     2.482589 Tokens per Sec:     7594, Lr: 0.000300\n",
      "2020-02-20 12:42:57,040 Epoch   5 Step:     6400 Batch Loss:     2.914473 Tokens per Sec:     7581, Lr: 0.000300\n",
      "2020-02-20 12:43:25,492 Epoch   5 Step:     6500 Batch Loss:     2.596678 Tokens per Sec:     7547, Lr: 0.000300\n",
      "2020-02-20 12:43:54,212 Epoch   5 Step:     6600 Batch Loss:     2.626672 Tokens per Sec:     7422, Lr: 0.000300\n",
      "2020-02-20 12:44:22,613 Epoch   5 Step:     6700 Batch Loss:     2.924554 Tokens per Sec:     7495, Lr: 0.000300\n",
      "2020-02-20 12:44:51,306 Epoch   5 Step:     6800 Batch Loss:     2.206463 Tokens per Sec:     7455, Lr: 0.000300\n",
      "2020-02-20 12:45:20,424 Epoch   5 Step:     6900 Batch Loss:     2.519378 Tokens per Sec:     7674, Lr: 0.000300\n",
      "2020-02-20 12:45:49,385 Epoch   5 Step:     7000 Batch Loss:     2.784849 Tokens per Sec:     7641, Lr: 0.000300\n",
      "2020-02-20 12:47:17,551 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:47:17,552 Saving new checkpoint.\n",
      "2020-02-20 12:47:17,769 Example #0\n",
      "2020-02-20 12:47:17,769 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:47:17,769 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:47:17,769 \tHypothesis: Gweth ma Jehova dwaro ni giluw gima onego omi omi wabed gi paro mowinjore kuom Nyasaye .\n",
      "2020-02-20 12:47:17,769 Example #1\n",
      "2020-02-20 12:47:17,769 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:47:17,770 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:47:17,770 \tHypothesis: Bende , amor ahinya ni en gi luor ma ne an - go e lamo Nyasaye kendo ma en - go . ”\n",
      "2020-02-20 12:47:17,770 Example #2\n",
      "2020-02-20 12:47:17,770 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:47:17,770 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:47:17,770 \tHypothesis: Kuom adier , waparo matut kuom Ndiko nyalo konyowa ng’eyo kaka wanyalo bedo gi chuny mokuwe .\n",
      "2020-02-20 12:47:17,770 Example #3\n",
      "2020-02-20 12:47:17,770 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:47:17,770 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:47:17,770 \tHypothesis: Ne ooro weche ma ne okor e wi “ gik moko duto ma Nyasaye nowacho kuom janabi Isaya . ”\n",
      "2020-02-20 12:47:17,771 Validation result at epoch   5, step     7000: bleu:  11.96, loss: 60268.7461, ppl:  11.5326, duration: 88.3848s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 12:47:46,809 Epoch   5 Step:     7100 Batch Loss:     2.377353 Tokens per Sec:     7536, Lr: 0.000300\n",
      "2020-02-20 12:48:15,570 Epoch   5 Step:     7200 Batch Loss:     1.793351 Tokens per Sec:     7642, Lr: 0.000300\n",
      "2020-02-20 12:48:44,278 Epoch   5 Step:     7300 Batch Loss:     2.368781 Tokens per Sec:     7535, Lr: 0.000300\n",
      "2020-02-20 12:49:12,767 Epoch   5 Step:     7400 Batch Loss:     3.025920 Tokens per Sec:     7526, Lr: 0.000300\n",
      "2020-02-20 12:49:41,451 Epoch   5 Step:     7500 Batch Loss:     2.123299 Tokens per Sec:     7625, Lr: 0.000300\n",
      "2020-02-20 12:50:10,079 Epoch   5 Step:     7600 Batch Loss:     2.233503 Tokens per Sec:     7551, Lr: 0.000300\n",
      "2020-02-20 12:50:10,224 Epoch   5: total training loss 3924.77\n",
      "2020-02-20 12:50:10,224 EPOCH 6\n",
      "2020-02-20 12:50:38,666 Epoch   6 Step:     7700 Batch Loss:     2.196222 Tokens per Sec:     7480, Lr: 0.000300\n",
      "2020-02-20 12:51:07,374 Epoch   6 Step:     7800 Batch Loss:     2.038606 Tokens per Sec:     7539, Lr: 0.000300\n",
      "2020-02-20 12:51:36,278 Epoch   6 Step:     7900 Batch Loss:     2.617388 Tokens per Sec:     7576, Lr: 0.000300\n",
      "2020-02-20 12:52:04,641 Epoch   6 Step:     8000 Batch Loss:     2.530022 Tokens per Sec:     7480, Lr: 0.000300\n",
      "2020-02-20 12:53:32,809 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:53:32,810 Saving new checkpoint.\n",
      "2020-02-20 12:53:33,028 Example #0\n",
      "2020-02-20 12:53:33,029 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:53:33,029 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:53:33,029 \tHypothesis: Ginyiso ni giikore luwo gima Jehova dwaro ni mondo omi dhano obed gi paro mowinjore .\n",
      "2020-02-20 12:53:33,029 Example #1\n",
      "2020-02-20 12:53:33,029 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:53:33,029 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:53:33,029 \tHypothesis: Bende , achiwo luor ne wuonwa kaka wuoro kendo ni en gi winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 12:53:33,029 Example #2\n",
      "2020-02-20 12:53:33,029 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:53:33,029 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:53:33,029 \tHypothesis: Kuom adier , sama wawuoyo e Ndiko e yo ma kamano nyalo miyo ng’ato obed gi chuny maler .\n",
      "2020-02-20 12:53:33,030 Example #3\n",
      "2020-02-20 12:53:33,030 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:53:33,030 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:53:33,030 \tHypothesis: Ne okoro weche ma ne okor e wi “ gik moko duto ma Nyasaye nowacho kokalo kuom janabi maler , ma ne okor e kinde ma ne okor . ”\n",
      "2020-02-20 12:53:33,030 Validation result at epoch   6, step     8000: bleu:  12.98, loss: 58475.8125, ppl:  10.7235, duration: 88.3884s\n",
      "2020-02-20 12:54:01,854 Epoch   6 Step:     8100 Batch Loss:     2.635342 Tokens per Sec:     7650, Lr: 0.000300\n",
      "2020-02-20 12:54:30,728 Epoch   6 Step:     8200 Batch Loss:     2.719745 Tokens per Sec:     7628, Lr: 0.000300\n",
      "2020-02-20 12:54:59,789 Epoch   6 Step:     8300 Batch Loss:     2.731328 Tokens per Sec:     7610, Lr: 0.000300\n",
      "2020-02-20 12:55:28,959 Epoch   6 Step:     8400 Batch Loss:     2.633184 Tokens per Sec:     7656, Lr: 0.000300\n",
      "2020-02-20 12:55:57,197 Epoch   6 Step:     8500 Batch Loss:     2.693359 Tokens per Sec:     7364, Lr: 0.000300\n",
      "2020-02-20 12:56:26,347 Epoch   6 Step:     8600 Batch Loss:     2.647988 Tokens per Sec:     7670, Lr: 0.000300\n",
      "2020-02-20 12:56:55,051 Epoch   6 Step:     8700 Batch Loss:     2.646337 Tokens per Sec:     7531, Lr: 0.000300\n",
      "2020-02-20 12:57:24,148 Epoch   6 Step:     8800 Batch Loss:     1.519602 Tokens per Sec:     7492, Lr: 0.000300\n",
      "2020-02-20 12:57:53,096 Epoch   6 Step:     8900 Batch Loss:     2.236732 Tokens per Sec:     7625, Lr: 0.000300\n",
      "2020-02-20 12:58:22,081 Epoch   6 Step:     9000 Batch Loss:     2.326519 Tokens per Sec:     7572, Lr: 0.000300\n",
      "2020-02-20 12:59:50,300 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 12:59:50,300 Saving new checkpoint.\n",
      "2020-02-20 12:59:50,516 Example #0\n",
      "2020-02-20 12:59:50,516 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 12:59:50,517 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 12:59:50,517 \tHypothesis: Gombo ma gichiwo ne Jehova dwaro ni otim gik moko duto ma Nyasaye dwaro ni otim .\n",
      "2020-02-20 12:59:50,517 Example #1\n",
      "2020-02-20 12:59:50,517 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 12:59:50,517 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 12:59:50,517 \tHypothesis: Bende , en gima ber mondo awinj ka wuon mare kendo omakore gi Nyasaye . ”\n",
      "2020-02-20 12:59:50,517 Example #2\n",
      "2020-02-20 12:59:50,517 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 12:59:50,517 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 12:59:50,517 \tHypothesis: Ket pachwa e Ndiko , nyalo bedo ni ng’ato nyalo ng’eyo tiend weche manie Ndiko .\n",
      "2020-02-20 12:59:50,517 Example #3\n",
      "2020-02-20 12:59:50,518 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 12:59:50,518 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 12:59:50,518 \tHypothesis: Ne ooro weche ma ne okor e wi “ gik moko duto ma Nyasaye nowacho kokalo kuom janabi maler . ”\n",
      "2020-02-20 12:59:50,518 Validation result at epoch   6, step     9000: bleu:  13.53, loss: 57032.0273, ppl:  10.1134, duration: 88.4363s\n",
      "2020-02-20 13:00:19,367 Epoch   6 Step:     9100 Batch Loss:     2.202389 Tokens per Sec:     7487, Lr: 0.000300\n",
      "2020-02-20 13:00:26,260 Epoch   6: total training loss 3733.66\n",
      "2020-02-20 13:00:26,260 EPOCH 7\n",
      "2020-02-20 13:00:48,498 Epoch   7 Step:     9200 Batch Loss:     1.776751 Tokens per Sec:     7591, Lr: 0.000300\n",
      "2020-02-20 13:01:17,676 Epoch   7 Step:     9300 Batch Loss:     2.848773 Tokens per Sec:     7698, Lr: 0.000300\n",
      "2020-02-20 13:01:46,183 Epoch   7 Step:     9400 Batch Loss:     2.161055 Tokens per Sec:     7468, Lr: 0.000300\n",
      "2020-02-20 13:02:15,160 Epoch   7 Step:     9500 Batch Loss:     2.258479 Tokens per Sec:     7584, Lr: 0.000300\n",
      "2020-02-20 13:02:44,095 Epoch   7 Step:     9600 Batch Loss:     2.699597 Tokens per Sec:     7486, Lr: 0.000300\n",
      "2020-02-20 13:03:13,116 Epoch   7 Step:     9700 Batch Loss:     2.499038 Tokens per Sec:     7706, Lr: 0.000300\n",
      "2020-02-20 13:03:41,895 Epoch   7 Step:     9800 Batch Loss:     2.388414 Tokens per Sec:     7500, Lr: 0.000300\n",
      "2020-02-20 13:04:10,344 Epoch   7 Step:     9900 Batch Loss:     1.981571 Tokens per Sec:     7464, Lr: 0.000300\n",
      "2020-02-20 13:04:39,375 Epoch   7 Step:    10000 Batch Loss:     2.537346 Tokens per Sec:     7635, Lr: 0.000300\n",
      "2020-02-20 13:06:07,385 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:06:07,385 Saving new checkpoint.\n",
      "2020-02-20 13:06:07,601 Example #0\n",
      "2020-02-20 13:06:07,601 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:06:07,601 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:06:07,601 \tHypothesis: Ginyalo yie luwo gima Jehova dwaro ni watim mondo omi wabed joma kare .\n",
      "2020-02-20 13:06:07,602 Example #1\n",
      "2020-02-20 13:06:07,602 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:06:07,602 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:06:07,602 \tHypothesis: Bende , amiyo abedo gi luor kaka wuoro kod wuonwa . ”\n",
      "2020-02-20 13:06:07,602 Example #2\n",
      "2020-02-20 13:06:07,602 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:06:07,602 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:06:07,602 \tHypothesis: Ng’e ni mondo wamed ler e Ndiko , nyalo bedo gi teko mar ng’eyo kaka chuny ng’ato nyalo nenore e chunye .\n",
      "2020-02-20 13:06:07,602 Example #3\n",
      "2020-02-20 13:06:07,603 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:06:07,603 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:06:07,603 \tHypothesis: Ne ondiko ni “ giduto ma Nyasaye ne ochiwo , ma ne owachone kuom janabi maler . ”\n",
      "2020-02-20 13:06:07,603 Validation result at epoch   7, step    10000: bleu:  15.22, loss: 54667.5508, ppl:   9.1883, duration: 88.2267s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 13:06:36,315 Epoch   7 Step:    10100 Batch Loss:     2.194897 Tokens per Sec:     7569, Lr: 0.000300\n",
      "2020-02-20 13:07:04,902 Epoch   7 Step:    10200 Batch Loss:     2.210200 Tokens per Sec:     7497, Lr: 0.000300\n",
      "2020-02-20 13:07:33,669 Epoch   7 Step:    10300 Batch Loss:     1.960394 Tokens per Sec:     7572, Lr: 0.000300\n",
      "2020-02-20 13:08:02,569 Epoch   7 Step:    10400 Batch Loss:     2.159416 Tokens per Sec:     7490, Lr: 0.000300\n",
      "2020-02-20 13:08:31,622 Epoch   7 Step:    10500 Batch Loss:     2.274217 Tokens per Sec:     7653, Lr: 0.000300\n",
      "2020-02-20 13:09:00,132 Epoch   7 Step:    10600 Batch Loss:     2.411942 Tokens per Sec:     7477, Lr: 0.000300\n",
      "2020-02-20 13:09:13,505 Epoch   7: total training loss 3569.71\n",
      "2020-02-20 13:09:13,506 EPOCH 8\n",
      "2020-02-20 13:09:28,869 Epoch   8 Step:    10700 Batch Loss:     2.155770 Tokens per Sec:     7371, Lr: 0.000300\n",
      "2020-02-20 13:09:57,690 Epoch   8 Step:    10800 Batch Loss:     1.774834 Tokens per Sec:     7595, Lr: 0.000300\n",
      "2020-02-20 13:10:26,338 Epoch   8 Step:    10900 Batch Loss:     2.219854 Tokens per Sec:     7511, Lr: 0.000300\n",
      "2020-02-20 13:10:54,983 Epoch   8 Step:    11000 Batch Loss:     2.169687 Tokens per Sec:     7522, Lr: 0.000300\n",
      "2020-02-20 13:12:23,148 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:12:23,148 Saving new checkpoint.\n",
      "2020-02-20 13:12:23,367 Example #0\n",
      "2020-02-20 13:12:23,368 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:12:23,368 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:12:23,368 \tHypothesis: Giikore luwo gima Jehova dwaro ni mondo omi dhano obed gi paro makare .\n",
      "2020-02-20 13:12:23,368 Example #1\n",
      "2020-02-20 13:12:23,368 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:12:23,368 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:12:23,368 \tHypothesis: Bende , ahero miyo wuonwa luor kendo e wi Nyasaye . ”\n",
      "2020-02-20 13:12:23,368 Example #2\n",
      "2020-02-20 13:12:23,369 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:12:23,369 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:12:23,369 \tHypothesis: Bedo gi ler e Ndiko e yo matut , nyalo bedo ni iwuoyo kuom gik ma nyalo miyo ng’ato obed gi teko e chunye .\n",
      "2020-02-20 13:12:23,369 Example #3\n",
      "2020-02-20 13:12:23,369 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:12:23,369 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:12:23,369 \tHypothesis: Ne ondiko ni “ gik moko duto ma Nyasaye notiyogo e kinde ma ne en rawera . ”\n",
      "2020-02-20 13:12:23,369 Validation result at epoch   8, step    11000: bleu:  15.06, loss: 53643.9844, ppl:   8.8145, duration: 88.3858s\n",
      "2020-02-20 13:12:52,436 Epoch   8 Step:    11100 Batch Loss:     2.169235 Tokens per Sec:     7743, Lr: 0.000300\n",
      "2020-02-20 13:13:21,121 Epoch   8 Step:    11200 Batch Loss:     2.335445 Tokens per Sec:     7505, Lr: 0.000300\n",
      "2020-02-20 13:13:50,408 Epoch   8 Step:    11300 Batch Loss:     2.389638 Tokens per Sec:     7737, Lr: 0.000300\n",
      "2020-02-20 13:14:19,395 Epoch   8 Step:    11400 Batch Loss:     2.325415 Tokens per Sec:     7615, Lr: 0.000300\n",
      "2020-02-20 13:14:48,605 Epoch   8 Step:    11500 Batch Loss:     2.009950 Tokens per Sec:     7586, Lr: 0.000300\n",
      "2020-02-20 13:15:17,295 Epoch   8 Step:    11600 Batch Loss:     2.069922 Tokens per Sec:     7589, Lr: 0.000300\n",
      "2020-02-20 13:15:46,339 Epoch   8 Step:    11700 Batch Loss:     2.321386 Tokens per Sec:     7531, Lr: 0.000300\n",
      "2020-02-20 13:16:14,865 Epoch   8 Step:    11800 Batch Loss:     2.287055 Tokens per Sec:     7496, Lr: 0.000300\n",
      "2020-02-20 13:16:43,994 Epoch   8 Step:    11900 Batch Loss:     2.203277 Tokens per Sec:     7641, Lr: 0.000300\n",
      "2020-02-20 13:17:12,778 Epoch   8 Step:    12000 Batch Loss:     2.154271 Tokens per Sec:     7618, Lr: 0.000300\n",
      "2020-02-20 13:18:40,903 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:18:40,903 Saving new checkpoint.\n",
      "2020-02-20 13:18:41,120 Example #0\n",
      "2020-02-20 13:18:41,120 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:18:41,120 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:18:41,120 \tHypothesis: Giikore luwo chike ma Jehova dwaro ni mondo omi giluw chike mag Nyasaye .\n",
      "2020-02-20 13:18:41,120 Example #1\n",
      "2020-02-20 13:18:41,120 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:18:41,121 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:18:41,121 \tHypothesis: Bende , omiya luor momedore mar bedo wuon - gi kendo tiyo ne Nyasaye . ”\n",
      "2020-02-20 13:18:41,121 Example #2\n",
      "2020-02-20 13:18:41,121 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:18:41,121 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:18:41,121 \tHypothesis: Ket pachwa kuom gik ma Ndiko wacho nyalo bedo ni ok wanyal ng’eyo ni ok wanyal ng’eyo kaka chunywa dwaro .\n",
      "2020-02-20 13:18:41,121 Example #3\n",
      "2020-02-20 13:18:41,121 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:18:41,121 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:18:41,121 \tHypothesis: Ne ondiko weche ma ne okor e wi “ gik moko duto ma Nyasaye nowacho kokalo kuom janabi maler mar roho maler . ”\n",
      "2020-02-20 13:18:41,122 Validation result at epoch   8, step    12000: bleu:  15.88, loss: 52487.8945, ppl:   8.4107, duration: 88.3429s\n",
      "2020-02-20 13:19:09,691 Epoch   8 Step:    12100 Batch Loss:     2.467495 Tokens per Sec:     7483, Lr: 0.000300\n",
      "2020-02-20 13:19:28,266 Epoch   8: total training loss 3424.48\n",
      "2020-02-20 13:19:28,267 EPOCH 9\n",
      "2020-02-20 13:19:38,825 Epoch   9 Step:    12200 Batch Loss:     2.104245 Tokens per Sec:     7420, Lr: 0.000300\n",
      "2020-02-20 13:20:07,670 Epoch   9 Step:    12300 Batch Loss:     2.073637 Tokens per Sec:     7726, Lr: 0.000300\n",
      "2020-02-20 13:20:36,639 Epoch   9 Step:    12400 Batch Loss:     2.399733 Tokens per Sec:     7617, Lr: 0.000300\n",
      "2020-02-20 13:21:05,343 Epoch   9 Step:    12500 Batch Loss:     2.354544 Tokens per Sec:     7494, Lr: 0.000300\n",
      "2020-02-20 13:21:34,006 Epoch   9 Step:    12600 Batch Loss:     1.874124 Tokens per Sec:     7479, Lr: 0.000300\n",
      "2020-02-20 13:22:02,608 Epoch   9 Step:    12700 Batch Loss:     2.101764 Tokens per Sec:     7585, Lr: 0.000300\n",
      "2020-02-20 13:22:31,258 Epoch   9 Step:    12800 Batch Loss:     1.753699 Tokens per Sec:     7707, Lr: 0.000300\n",
      "2020-02-20 13:23:00,161 Epoch   9 Step:    12900 Batch Loss:     2.401218 Tokens per Sec:     7578, Lr: 0.000300\n",
      "2020-02-20 13:23:29,053 Epoch   9 Step:    13000 Batch Loss:     1.602911 Tokens per Sec:     7646, Lr: 0.000300\n",
      "2020-02-20 13:24:57,106 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:24:57,106 Saving new checkpoint.\n",
      "2020-02-20 13:24:57,320 Example #0\n",
      "2020-02-20 13:24:57,320 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:24:57,320 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:24:57,320 \tHypothesis: Luwo chikene ne Jehova , dwaro ni mondo omi Jehova obed gi paro mowinjore .\n",
      "2020-02-20 13:24:57,320 Example #1\n",
      "2020-02-20 13:24:57,321 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:24:57,321 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:24:57,321 \tHypothesis: Bende , amiyo abedo gi luor kaka wuoro kendo e wi winjruokna gi Nyasaye . ”\n",
      "2020-02-20 13:24:57,321 Example #2\n",
      "2020-02-20 13:24:57,321 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:24:57,321 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:24:57,321 \tHypothesis: Ket pachi kuom gik ma Ndiko wacho e wi Muma , nyalo bedo gima nyiso ni en gi teko mar ng’eyo gik ma nyalo miyo ng’ato otim .\n",
      "2020-02-20 13:24:57,321 Example #3\n",
      "2020-02-20 13:24:57,322 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:24:57,322 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:24:57,322 \tHypothesis: Ne ondiko e wi “ gik moko duto ma Nyasaye nowacho kokalo kuom janabi maler , ma ne en janabi maler . ”\n",
      "2020-02-20 13:24:57,322 Validation result at epoch   9, step    13000: bleu:  16.27, loss: 51491.7305, ppl:   8.0775, duration: 88.2683s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 13:25:26,199 Epoch   9 Step:    13100 Batch Loss:     1.966934 Tokens per Sec:     7477, Lr: 0.000300\n",
      "2020-02-20 13:25:54,608 Epoch   9 Step:    13200 Batch Loss:     2.030701 Tokens per Sec:     7424, Lr: 0.000300\n",
      "2020-02-20 13:26:23,230 Epoch   9 Step:    13300 Batch Loss:     2.384551 Tokens per Sec:     7609, Lr: 0.000300\n",
      "2020-02-20 13:26:52,018 Epoch   9 Step:    13400 Batch Loss:     2.075567 Tokens per Sec:     7556, Lr: 0.000300\n",
      "2020-02-20 13:27:20,908 Epoch   9 Step:    13500 Batch Loss:     2.342129 Tokens per Sec:     7744, Lr: 0.000300\n",
      "2020-02-20 13:27:50,110 Epoch   9 Step:    13600 Batch Loss:     2.523449 Tokens per Sec:     7784, Lr: 0.000300\n",
      "2020-02-20 13:28:13,814 Epoch   9: total training loss 3318.59\n",
      "2020-02-20 13:28:13,814 EPOCH 10\n",
      "2020-02-20 13:28:19,276 Epoch  10 Step:    13700 Batch Loss:     2.016206 Tokens per Sec:     7363, Lr: 0.000300\n",
      "2020-02-20 13:28:48,178 Epoch  10 Step:    13800 Batch Loss:     2.267726 Tokens per Sec:     7665, Lr: 0.000300\n",
      "2020-02-20 13:29:16,816 Epoch  10 Step:    13900 Batch Loss:     1.974838 Tokens per Sec:     7502, Lr: 0.000300\n",
      "2020-02-20 13:29:45,897 Epoch  10 Step:    14000 Batch Loss:     2.123787 Tokens per Sec:     7670, Lr: 0.000300\n",
      "2020-02-20 13:31:14,015 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:31:14,015 Saving new checkpoint.\n",
      "2020-02-20 13:31:14,234 Example #0\n",
      "2020-02-20 13:31:14,234 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:31:14,234 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:31:14,234 \tHypothesis: Ka giikore winjo wach Jehova , onego onyis ni loch mar Nyasaye ok nyal timo gik moko .\n",
      "2020-02-20 13:31:14,234 Example #1\n",
      "2020-02-20 13:31:14,235 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:31:14,235 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:31:14,235 \tHypothesis: Bende , amiyo amiye luor kaka wuonwa koda e wi Nyasaye . ”\n",
      "2020-02-20 13:31:14,235 Example #2\n",
      "2020-02-20 13:31:14,235 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:31:14,235 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:31:14,235 \tHypothesis: Bedo gi ler e yo matut kuom gima Ndiko wacho nyalo bedo gi teko e fwenyo ni chuny ng’ato nyalo fwenyo .\n",
      "2020-02-20 13:31:14,235 Example #3\n",
      "2020-02-20 13:31:14,235 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:31:14,236 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:31:14,236 \tHypothesis: Ne ondiko e wi “ gik moko duto ma Nyasaye ne wuoyo kuome e wi janabi maler . ”\n",
      "2020-02-20 13:31:14,236 Validation result at epoch  10, step    14000: bleu:  17.06, loss: 50492.8320, ppl:   7.7567, duration: 88.3380s\n",
      "2020-02-20 13:31:42,985 Epoch  10 Step:    14100 Batch Loss:     2.337510 Tokens per Sec:     7557, Lr: 0.000300\n",
      "2020-02-20 13:32:11,735 Epoch  10 Step:    14200 Batch Loss:     2.451478 Tokens per Sec:     7623, Lr: 0.000300\n",
      "2020-02-20 13:32:40,354 Epoch  10 Step:    14300 Batch Loss:     2.118098 Tokens per Sec:     7528, Lr: 0.000300\n",
      "2020-02-20 13:33:09,327 Epoch  10 Step:    14400 Batch Loss:     2.292750 Tokens per Sec:     7625, Lr: 0.000300\n",
      "2020-02-20 13:33:38,332 Epoch  10 Step:    14500 Batch Loss:     2.262671 Tokens per Sec:     7602, Lr: 0.000300\n",
      "2020-02-20 13:34:07,505 Epoch  10 Step:    14600 Batch Loss:     1.570102 Tokens per Sec:     7724, Lr: 0.000300\n",
      "2020-02-20 13:34:36,435 Epoch  10 Step:    14700 Batch Loss:     2.276570 Tokens per Sec:     7625, Lr: 0.000300\n",
      "2020-02-20 13:35:05,400 Epoch  10 Step:    14800 Batch Loss:     2.175345 Tokens per Sec:     7543, Lr: 0.000300\n",
      "2020-02-20 13:35:34,127 Epoch  10 Step:    14900 Batch Loss:     2.063216 Tokens per Sec:     7544, Lr: 0.000300\n",
      "2020-02-20 13:36:02,679 Epoch  10 Step:    15000 Batch Loss:     2.254591 Tokens per Sec:     7480, Lr: 0.000300\n",
      "2020-02-20 13:37:30,781 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:37:30,781 Saving new checkpoint.\n",
      "2020-02-20 13:37:30,997 Example #0\n",
      "2020-02-20 13:37:30,998 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:37:30,998 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:37:30,998 \tHypothesis: Chiwruok margi giwegi gi winjo gima Jehova dwaro ni watim , onego onyis ni loch mar Nyasaye en gima owinjore .\n",
      "2020-02-20 13:37:30,998 Example #1\n",
      "2020-02-20 13:37:30,998 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:37:30,998 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:37:30,998 \tHypothesis: Bende , amiyo awinj ka an gi luor , kendo ni an gi winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 13:37:30,998 Example #2\n",
      "2020-02-20 13:37:30,998 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:37:30,999 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:37:30,999 \tHypothesis: Ng’eyo ni wan wawegi warieyo gik ma Ndiko wacho nyalo bedo gima duong ’ ahinya e fwenyo kaka chuny ng’ato nyalo bedo gi chuny mokuwe .\n",
      "2020-02-20 13:37:30,999 Example #3\n",
      "2020-02-20 13:37:30,999 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:37:30,999 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:37:30,999 \tHypothesis: Nondiko weche ma ne okor e wi “ gik moko duto ma Nyasaye ne wuoyo kuome kokalo kuom janabi maler . ”\n",
      "2020-02-20 13:37:30,999 Validation result at epoch  10, step    15000: bleu:  18.30, loss: 49596.8750, ppl:   7.4798, duration: 88.3195s\n",
      "2020-02-20 13:37:59,772 Epoch  10 Step:    15100 Batch Loss:     2.018469 Tokens per Sec:     7621, Lr: 0.000300\n",
      "2020-02-20 13:38:27,994 Epoch  10: total training loss 3217.65\n",
      "2020-02-20 13:38:27,994 EPOCH 11\n",
      "2020-02-20 13:38:28,817 Epoch  11 Step:    15200 Batch Loss:     2.301722 Tokens per Sec:     6638, Lr: 0.000300\n",
      "2020-02-20 13:38:57,385 Epoch  11 Step:    15300 Batch Loss:     1.495556 Tokens per Sec:     7584, Lr: 0.000300\n",
      "2020-02-20 13:39:25,909 Epoch  11 Step:    15400 Batch Loss:     1.303177 Tokens per Sec:     7512, Lr: 0.000300\n",
      "2020-02-20 13:39:54,449 Epoch  11 Step:    15500 Batch Loss:     1.359732 Tokens per Sec:     7428, Lr: 0.000300\n",
      "2020-02-20 13:40:23,046 Epoch  11 Step:    15600 Batch Loss:     2.006906 Tokens per Sec:     7684, Lr: 0.000300\n",
      "2020-02-20 13:40:52,076 Epoch  11 Step:    15700 Batch Loss:     1.985547 Tokens per Sec:     7695, Lr: 0.000300\n",
      "2020-02-20 13:41:20,618 Epoch  11 Step:    15800 Batch Loss:     2.046363 Tokens per Sec:     7562, Lr: 0.000300\n",
      "2020-02-20 13:41:49,285 Epoch  11 Step:    15900 Batch Loss:     1.693688 Tokens per Sec:     7541, Lr: 0.000300\n",
      "2020-02-20 13:42:17,959 Epoch  11 Step:    16000 Batch Loss:     1.819900 Tokens per Sec:     7676, Lr: 0.000300\n",
      "2020-02-20 13:43:46,033 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:43:46,034 Saving new checkpoint.\n",
      "2020-02-20 13:43:46,251 Example #0\n",
      "2020-02-20 13:43:46,251 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:43:46,251 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:43:46,251 \tHypothesis: Chiwruok margi giwegi gi gik ma Jehova dwaro ni mondo onyis ni loch mar Nyasaye en gima owinjore .\n",
      "2020-02-20 13:43:46,252 Example #1\n",
      "2020-02-20 13:43:46,252 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:43:46,252 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:43:46,252 \tHypothesis: Bende , amiyo amiye luor kaka wuonwa koda winjruokna gi Nyasaye . ”\n",
      "2020-02-20 13:43:46,252 Example #2\n",
      "2020-02-20 13:43:46,252 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:43:46,252 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:43:46,252 \tHypothesis: Bedo gi ler e ler mar Ndiko , nyalo bedo gima nyiso ni ng’ato nigi teko mar ng’eyo chuny .\n",
      "2020-02-20 13:43:46,252 Example #3\n",
      "2020-02-20 13:43:46,253 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:43:46,253 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:43:46,253 \tHypothesis: Ne ondiko e wi “ gik moko duto ma Nyasaye ne wuoyo kuome e dhoge mag ruodhe , ” ma ne ondik e wi “ chiemo duto ma Nyasaye ne tiyogo e dhoge . ”\n",
      "2020-02-20 13:43:46,253 Validation result at epoch  11, step    16000: bleu:  17.64, loss: 49577.4414, ppl:   7.4739, duration: 88.2932s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 13:44:15,485 Epoch  11 Step:    16100 Batch Loss:     1.779898 Tokens per Sec:     7723, Lr: 0.000300\n",
      "2020-02-20 13:44:44,008 Epoch  11 Step:    16200 Batch Loss:     1.693069 Tokens per Sec:     7414, Lr: 0.000300\n",
      "2020-02-20 13:45:12,976 Epoch  11 Step:    16300 Batch Loss:     2.256828 Tokens per Sec:     7699, Lr: 0.000300\n",
      "2020-02-20 13:45:41,342 Epoch  11 Step:    16400 Batch Loss:     2.473014 Tokens per Sec:     7479, Lr: 0.000300\n",
      "2020-02-20 13:46:10,144 Epoch  11 Step:    16500 Batch Loss:     1.979323 Tokens per Sec:     7733, Lr: 0.000300\n",
      "2020-02-20 13:46:38,915 Epoch  11 Step:    16600 Batch Loss:     2.123430 Tokens per Sec:     7579, Lr: 0.000300\n",
      "2020-02-20 13:47:07,672 Epoch  11 Step:    16700 Batch Loss:     1.495558 Tokens per Sec:     7603, Lr: 0.000300\n",
      "2020-02-20 13:47:13,399 Epoch  11: total training loss 3163.59\n",
      "2020-02-20 13:47:13,400 EPOCH 12\n",
      "2020-02-20 13:47:36,731 Epoch  12 Step:    16800 Batch Loss:     1.909336 Tokens per Sec:     7616, Lr: 0.000300\n",
      "2020-02-20 13:48:05,468 Epoch  12 Step:    16900 Batch Loss:     2.172665 Tokens per Sec:     7641, Lr: 0.000300\n",
      "2020-02-20 13:48:34,176 Epoch  12 Step:    17000 Batch Loss:     1.714036 Tokens per Sec:     7586, Lr: 0.000300\n",
      "2020-02-20 13:50:02,251 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:50:02,251 Saving new checkpoint.\n",
      "2020-02-20 13:50:02,469 Example #0\n",
      "2020-02-20 13:50:02,470 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:50:02,470 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:50:02,470 \tHypothesis: Luwo chikene , dwaro ni Jehova ong’ad ni giluw chike Nyasaye .\n",
      "2020-02-20 13:50:02,470 Example #1\n",
      "2020-02-20 13:50:02,470 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:50:02,470 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:50:02,470 \tHypothesis: Bende , omiya luor momedore mar miye luor kaka wuonwa kod winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 13:50:02,470 Example #2\n",
      "2020-02-20 13:50:02,471 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:50:02,471 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:50:02,471 \tHypothesis: Kuom adier , ng’eyo tiend weche manie Ndiko nyalo bedo gima nyiso ni chuny ng’ato nyalo bedo maber .\n",
      "2020-02-20 13:50:02,471 Example #3\n",
      "2020-02-20 13:50:02,471 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:50:02,471 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:50:02,471 \tHypothesis: Nondiko ni “ gik moko duto mag Nyasaye ma Nyasaye nowuoyo kuome e wi janabi maler , ma ne en janabi maler . ”\n",
      "2020-02-20 13:50:02,472 Validation result at epoch  12, step    17000: bleu:  17.78, loss: 48477.7344, ppl:   7.1478, duration: 88.2948s\n",
      "2020-02-20 13:50:31,124 Epoch  12 Step:    17100 Batch Loss:     1.820889 Tokens per Sec:     7478, Lr: 0.000300\n",
      "2020-02-20 13:51:00,244 Epoch  12 Step:    17200 Batch Loss:     2.028999 Tokens per Sec:     7758, Lr: 0.000300\n",
      "2020-02-20 13:51:29,167 Epoch  12 Step:    17300 Batch Loss:     1.793971 Tokens per Sec:     7674, Lr: 0.000300\n",
      "2020-02-20 13:51:58,054 Epoch  12 Step:    17400 Batch Loss:     1.918590 Tokens per Sec:     7503, Lr: 0.000300\n",
      "2020-02-20 13:52:26,828 Epoch  12 Step:    17500 Batch Loss:     2.225218 Tokens per Sec:     7679, Lr: 0.000300\n",
      "2020-02-20 13:52:55,479 Epoch  12 Step:    17600 Batch Loss:     2.235759 Tokens per Sec:     7542, Lr: 0.000300\n",
      "2020-02-20 13:53:23,813 Epoch  12 Step:    17700 Batch Loss:     1.580759 Tokens per Sec:     7519, Lr: 0.000300\n",
      "2020-02-20 13:53:52,179 Epoch  12 Step:    17800 Batch Loss:     2.150620 Tokens per Sec:     7562, Lr: 0.000300\n",
      "2020-02-20 13:54:21,419 Epoch  12 Step:    17900 Batch Loss:     2.210498 Tokens per Sec:     7535, Lr: 0.000300\n",
      "2020-02-20 13:54:50,259 Epoch  12 Step:    18000 Batch Loss:     1.910157 Tokens per Sec:     7538, Lr: 0.000300\n",
      "2020-02-20 13:56:18,355 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 13:56:18,356 Saving new checkpoint.\n",
      "2020-02-20 13:56:18,572 Example #0\n",
      "2020-02-20 13:56:18,573 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 13:56:18,573 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 13:56:18,573 \tHypothesis: Luwo chikene , onego onyis ni Jehova ema dwaro ni ji obed gi paro mowinjore .\n",
      "2020-02-20 13:56:18,573 Example #1\n",
      "2020-02-20 13:56:18,573 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 13:56:18,573 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 13:56:18,573 \tHypothesis: Bende , omiya luor ahinya kuom miye luor kaka wuonwa koda wiye . ”\n",
      "2020-02-20 13:56:18,573 Example #2\n",
      "2020-02-20 13:56:18,574 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 13:56:18,574 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 13:56:18,574 \tHypothesis: Bedo gi chuny mar ler e weche manie Ndiko nyalo bedo gi teko mar ng’eyo gik ma chuny dwaro .\n",
      "2020-02-20 13:56:18,574 Example #3\n",
      "2020-02-20 13:56:18,574 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 13:56:18,574 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 13:56:18,574 \tHypothesis: Ne ondiko weche ma ne ondik e wi “ gik moko duto ma Nyasaye nowacho kokalo kuom janabi maler , ma ne en janabi maler . ”\n",
      "2020-02-20 13:56:18,574 Validation result at epoch  12, step    18000: bleu:  18.39, loss: 47651.5781, ppl:   6.9122, duration: 88.3147s\n",
      "2020-02-20 13:56:47,318 Epoch  12 Step:    18100 Batch Loss:     1.852726 Tokens per Sec:     7622, Lr: 0.000300\n",
      "2020-02-20 13:57:16,447 Epoch  12 Step:    18200 Batch Loss:     1.929286 Tokens per Sec:     7652, Lr: 0.000300\n",
      "2020-02-20 13:57:27,111 Epoch  12: total training loss 3082.26\n",
      "2020-02-20 13:57:27,112 EPOCH 13\n",
      "2020-02-20 13:57:45,315 Epoch  13 Step:    18300 Batch Loss:     1.983091 Tokens per Sec:     7587, Lr: 0.000300\n",
      "2020-02-20 13:58:13,985 Epoch  13 Step:    18400 Batch Loss:     1.967488 Tokens per Sec:     7543, Lr: 0.000300\n",
      "2020-02-20 13:58:42,869 Epoch  13 Step:    18500 Batch Loss:     1.554304 Tokens per Sec:     7553, Lr: 0.000300\n",
      "2020-02-20 13:59:11,558 Epoch  13 Step:    18600 Batch Loss:     2.260455 Tokens per Sec:     7531, Lr: 0.000300\n",
      "2020-02-20 13:59:40,090 Epoch  13 Step:    18700 Batch Loss:     2.003440 Tokens per Sec:     7649, Lr: 0.000300\n",
      "2020-02-20 14:00:08,627 Epoch  13 Step:    18800 Batch Loss:     2.120882 Tokens per Sec:     7553, Lr: 0.000300\n",
      "2020-02-20 14:00:37,738 Epoch  13 Step:    18900 Batch Loss:     2.058761 Tokens per Sec:     7696, Lr: 0.000300\n",
      "2020-02-20 14:01:06,400 Epoch  13 Step:    19000 Batch Loss:     2.203524 Tokens per Sec:     7554, Lr: 0.000300\n",
      "2020-02-20 14:02:34,245 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:02:34,246 Saving new checkpoint.\n",
      "2020-02-20 14:02:34,458 Example #0\n",
      "2020-02-20 14:02:34,459 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:02:34,459 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:02:34,459 \tHypothesis: Luwo chikene ne dwaro ni wanyis ni chike Nyasaye ema owinjore .\n",
      "2020-02-20 14:02:34,459 Example #1\n",
      "2020-02-20 14:02:34,459 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:02:34,459 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:02:34,459 \tHypothesis: Bende , omiya luor momedore e miyo amiye luor kaka wuoro kendo e wi Nyasaye . ”\n",
      "2020-02-20 14:02:34,459 Example #2\n",
      "2020-02-20 14:02:34,460 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:02:34,460 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:02:34,460 \tHypothesis: Ng’e ni mondo waler e ler mar Ndiko , wanyalo fwenyo ni chuny ng’ato nyalo fwenyo .\n",
      "2020-02-20 14:02:34,460 Example #3\n",
      "2020-02-20 14:02:34,460 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:02:34,460 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:02:34,460 \tHypothesis: Nondiko weche ma ne ondik e wi “ gik moko duto ma Nyasaye nowuoyo kuome e dho - Hibrania machon . ”\n",
      "2020-02-20 14:02:34,460 Validation result at epoch  13, step    19000: bleu:  18.75, loss: 47321.0898, ppl:   6.8201, duration: 88.0598s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 14:03:03,626 Epoch  13 Step:    19100 Batch Loss:     2.026862 Tokens per Sec:     7638, Lr: 0.000300\n",
      "2020-02-20 14:03:32,593 Epoch  13 Step:    19200 Batch Loss:     2.185729 Tokens per Sec:     7646, Lr: 0.000300\n",
      "2020-02-20 14:04:01,428 Epoch  13 Step:    19300 Batch Loss:     1.903270 Tokens per Sec:     7637, Lr: 0.000300\n",
      "2020-02-20 14:04:29,793 Epoch  13 Step:    19400 Batch Loss:     2.150892 Tokens per Sec:     7482, Lr: 0.000300\n",
      "2020-02-20 14:04:58,275 Epoch  13 Step:    19500 Batch Loss:     1.806008 Tokens per Sec:     7577, Lr: 0.000300\n",
      "2020-02-20 14:05:27,178 Epoch  13 Step:    19600 Batch Loss:     1.681479 Tokens per Sec:     7557, Lr: 0.000300\n",
      "2020-02-20 14:05:55,988 Epoch  13 Step:    19700 Batch Loss:     1.494733 Tokens per Sec:     7634, Lr: 0.000300\n",
      "2020-02-20 14:06:12,684 Epoch  13: total training loss 3037.68\n",
      "2020-02-20 14:06:12,684 EPOCH 14\n",
      "2020-02-20 14:06:24,718 Epoch  14 Step:    19800 Batch Loss:     1.871077 Tokens per Sec:     7269, Lr: 0.000300\n",
      "2020-02-20 14:06:53,463 Epoch  14 Step:    19900 Batch Loss:     1.997911 Tokens per Sec:     7417, Lr: 0.000300\n",
      "2020-02-20 14:07:22,548 Epoch  14 Step:    20000 Batch Loss:     2.252568 Tokens per Sec:     7592, Lr: 0.000300\n",
      "2020-02-20 14:08:50,622 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:08:50,623 Saving new checkpoint.\n",
      "2020-02-20 14:08:50,839 Example #0\n",
      "2020-02-20 14:08:50,839 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:08:50,840 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:08:50,840 \tHypothesis: Luwo chike Jehova , onego onyis ni chike Nyasaye ok owinjore .\n",
      "2020-02-20 14:08:50,840 Example #1\n",
      "2020-02-20 14:08:50,840 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:08:50,840 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:08:50,840 \tHypothesis: Bende , amiyo amiye luor ne wuonwa koda winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 14:08:50,840 Example #2\n",
      "2020-02-20 14:08:50,840 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:08:50,840 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:08:50,840 \tHypothesis: Ng’eyo ni Ndiko wacho adier nyalo bedo gi teko mar ng’eyo chuny .\n",
      "2020-02-20 14:08:50,841 Example #3\n",
      "2020-02-20 14:08:50,841 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:08:50,841 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:08:50,841 \tHypothesis: Ne ondiko ni “ weche duto mag Nyasaye ma Nyasaye nowuoyo kuome e dhoge machon . ”\n",
      "2020-02-20 14:08:50,841 Validation result at epoch  14, step    20000: bleu:  19.04, loss: 46756.9531, ppl:   6.6658, duration: 88.2920s\n",
      "2020-02-20 14:09:19,810 Epoch  14 Step:    20100 Batch Loss:     2.208756 Tokens per Sec:     7672, Lr: 0.000300\n",
      "2020-02-20 14:09:48,733 Epoch  14 Step:    20200 Batch Loss:     2.163957 Tokens per Sec:     7615, Lr: 0.000300\n",
      "2020-02-20 14:10:17,489 Epoch  14 Step:    20300 Batch Loss:     2.115371 Tokens per Sec:     7475, Lr: 0.000300\n",
      "2020-02-20 14:10:46,010 Epoch  14 Step:    20400 Batch Loss:     1.851070 Tokens per Sec:     7497, Lr: 0.000300\n",
      "2020-02-20 14:11:15,011 Epoch  14 Step:    20500 Batch Loss:     1.853435 Tokens per Sec:     7626, Lr: 0.000300\n",
      "2020-02-20 14:11:43,895 Epoch  14 Step:    20600 Batch Loss:     1.965919 Tokens per Sec:     7583, Lr: 0.000300\n",
      "2020-02-20 14:12:12,501 Epoch  14 Step:    20700 Batch Loss:     2.373045 Tokens per Sec:     7564, Lr: 0.000300\n",
      "2020-02-20 14:12:41,520 Epoch  14 Step:    20800 Batch Loss:     2.145957 Tokens per Sec:     7713, Lr: 0.000300\n",
      "2020-02-20 14:13:10,329 Epoch  14 Step:    20900 Batch Loss:     2.221987 Tokens per Sec:     7523, Lr: 0.000300\n",
      "2020-02-20 14:13:39,002 Epoch  14 Step:    21000 Batch Loss:     2.147277 Tokens per Sec:     7573, Lr: 0.000300\n",
      "2020-02-20 14:15:07,091 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:15:07,092 Saving new checkpoint.\n",
      "2020-02-20 14:15:07,309 Example #0\n",
      "2020-02-20 14:15:07,310 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:15:07,310 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:15:07,310 \tHypothesis: Chiwruok margi giwegi gi winjo wach ma Jehova dwaro ni mondo onyis ni chike mag Nyasaye ok owinjore .\n",
      "2020-02-20 14:15:07,310 Example #1\n",
      "2020-02-20 14:15:07,310 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:15:07,310 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:15:07,310 \tHypothesis: Bende , amiyo amiye luor ne wuonwa koda e wi Nyasaye . ”\n",
      "2020-02-20 14:15:07,310 Example #2\n",
      "2020-02-20 14:15:07,311 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:15:07,311 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:15:07,311 \tHypothesis: Ng’eyo ni Muma wacho adier nyalo bedo gima duong ’ ahinya e chuny ng’ato .\n",
      "2020-02-20 14:15:07,311 Example #3\n",
      "2020-02-20 14:15:07,311 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:15:07,311 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:15:07,311 \tHypothesis: Nondike e wi “ weche duto mag Nyasaye ma Nyasaye nowuoyo kuome e dhoge machon . ”\n",
      "2020-02-20 14:15:07,311 Validation result at epoch  14, step    21000: bleu:  19.73, loss: 46212.3164, ppl:   6.5201, duration: 88.3087s\n",
      "2020-02-20 14:15:35,732 Epoch  14 Step:    21100 Batch Loss:     1.614484 Tokens per Sec:     7533, Lr: 0.000300\n",
      "2020-02-20 14:16:04,341 Epoch  14 Step:    21200 Batch Loss:     1.654272 Tokens per Sec:     7611, Lr: 0.000300\n",
      "2020-02-20 14:16:27,989 Epoch  14: total training loss 2986.49\n",
      "2020-02-20 14:16:27,989 EPOCH 15\n",
      "2020-02-20 14:16:33,131 Epoch  15 Step:    21300 Batch Loss:     1.952488 Tokens per Sec:     7229, Lr: 0.000300\n",
      "2020-02-20 14:17:01,804 Epoch  15 Step:    21400 Batch Loss:     1.934405 Tokens per Sec:     7552, Lr: 0.000300\n",
      "2020-02-20 14:17:30,627 Epoch  15 Step:    21500 Batch Loss:     1.955800 Tokens per Sec:     7703, Lr: 0.000300\n",
      "2020-02-20 14:17:59,378 Epoch  15 Step:    21600 Batch Loss:     2.198103 Tokens per Sec:     7598, Lr: 0.000300\n",
      "2020-02-20 14:18:28,247 Epoch  15 Step:    21700 Batch Loss:     2.100889 Tokens per Sec:     7703, Lr: 0.000300\n",
      "2020-02-20 14:18:57,290 Epoch  15 Step:    21800 Batch Loss:     2.066184 Tokens per Sec:     7612, Lr: 0.000300\n",
      "2020-02-20 14:19:25,967 Epoch  15 Step:    21900 Batch Loss:     1.609360 Tokens per Sec:     7526, Lr: 0.000300\n",
      "2020-02-20 14:19:54,940 Epoch  15 Step:    22000 Batch Loss:     1.831302 Tokens per Sec:     7512, Lr: 0.000300\n",
      "2020-02-20 14:21:23,047 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:21:23,048 Saving new checkpoint.\n",
      "2020-02-20 14:21:23,266 Example #0\n",
      "2020-02-20 14:21:23,267 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:21:23,267 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:21:23,267 \tHypothesis: Luwo chik ma Jehova dwaro ni mondo giluw , onego onyis ni loch mar Nyasaye ok nyal timo .\n",
      "2020-02-20 14:21:23,267 Example #1\n",
      "2020-02-20 14:21:23,267 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:21:23,267 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:21:23,267 \tHypothesis: Bende , mano miyo amiye luor kaka wuonwa kendo e wi Nyasaye . ”\n",
      "2020-02-20 14:21:23,268 Example #2\n",
      "2020-02-20 14:21:23,268 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:21:23,268 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:21:23,268 \tHypothesis: Ng’eyo ni Muma wacho adier nyalo bedo gima duong ’ ahinya e wang ’ chuny .\n",
      "2020-02-20 14:21:23,268 Example #3\n",
      "2020-02-20 14:21:23,268 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:21:23,268 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:21:23,268 \tHypothesis: Ne ondiko e wi “ weche duto mag Nyasaye ma Nyasaye ne wuoyo kuome e dhoge machon . ”\n",
      "2020-02-20 14:21:23,268 Validation result at epoch  15, step    22000: bleu:  19.07, loss: 46018.4766, ppl:   6.4690, duration: 88.3280s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 14:21:52,230 Epoch  15 Step:    22100 Batch Loss:     1.829914 Tokens per Sec:     7648, Lr: 0.000300\n",
      "2020-02-20 14:22:21,127 Epoch  15 Step:    22200 Batch Loss:     2.466875 Tokens per Sec:     7643, Lr: 0.000300\n",
      "2020-02-20 14:22:50,008 Epoch  15 Step:    22300 Batch Loss:     1.900896 Tokens per Sec:     7610, Lr: 0.000300\n",
      "2020-02-20 14:23:18,907 Epoch  15 Step:    22400 Batch Loss:     1.941167 Tokens per Sec:     7638, Lr: 0.000300\n",
      "2020-02-20 14:23:47,500 Epoch  15 Step:    22500 Batch Loss:     2.090586 Tokens per Sec:     7550, Lr: 0.000300\n",
      "2020-02-20 14:24:16,126 Epoch  15 Step:    22600 Batch Loss:     2.041277 Tokens per Sec:     7500, Lr: 0.000300\n",
      "2020-02-20 14:24:44,754 Epoch  15 Step:    22700 Batch Loss:     2.085501 Tokens per Sec:     7495, Lr: 0.000300\n",
      "2020-02-20 14:25:13,195 Epoch  15 Step:    22800 Batch Loss:     2.269855 Tokens per Sec:     7553, Lr: 0.000300\n",
      "2020-02-20 14:25:13,805 Epoch  15: total training loss 2935.06\n",
      "2020-02-20 14:25:13,806 EPOCH 16\n",
      "2020-02-20 14:25:42,081 Epoch  16 Step:    22900 Batch Loss:     1.973166 Tokens per Sec:     7561, Lr: 0.000300\n",
      "2020-02-20 14:26:10,852 Epoch  16 Step:    23000 Batch Loss:     2.282125 Tokens per Sec:     7458, Lr: 0.000300\n",
      "2020-02-20 14:27:39,022 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:27:39,022 Saving new checkpoint.\n",
      "2020-02-20 14:27:39,236 Example #0\n",
      "2020-02-20 14:27:39,236 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:27:39,237 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:27:39,237 \tHypothesis: Luwo chikene , onego onyis ni chike Nyasaye ema owinjore .\n",
      "2020-02-20 14:27:39,237 Example #1\n",
      "2020-02-20 14:27:39,237 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:27:39,237 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:27:39,237 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuonwa kod winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 14:27:39,237 Example #2\n",
      "2020-02-20 14:27:39,237 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:27:39,237 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:27:39,238 \tHypothesis: Bedo gi chuny mar ng’eyo kaka Muma wacho nyalo bedo gima duong ’ ahinya e fwenyo kaka chuny ng’ato nyalo bedo motegno .\n",
      "2020-02-20 14:27:39,238 Example #3\n",
      "2020-02-20 14:27:39,238 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:27:39,238 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:27:39,238 \tHypothesis: Noyudo osendiko e wi “ weche duto mag Nyasaye ma ne wuoyo kuom jonabi maler , ” ma ne en “ koth ndalo machon . ”\n",
      "2020-02-20 14:27:39,238 Validation result at epoch  16, step    23000: bleu:  19.68, loss: 45625.0352, ppl:   6.3666, duration: 88.3852s\n",
      "2020-02-20 14:28:07,895 Epoch  16 Step:    23100 Batch Loss:     1.623996 Tokens per Sec:     7526, Lr: 0.000300\n",
      "2020-02-20 14:28:36,968 Epoch  16 Step:    23200 Batch Loss:     2.171220 Tokens per Sec:     7538, Lr: 0.000300\n",
      "2020-02-20 14:29:06,002 Epoch  16 Step:    23300 Batch Loss:     1.812124 Tokens per Sec:     7634, Lr: 0.000300\n",
      "2020-02-20 14:29:34,646 Epoch  16 Step:    23400 Batch Loss:     1.823475 Tokens per Sec:     7604, Lr: 0.000300\n",
      "2020-02-20 14:30:03,539 Epoch  16 Step:    23500 Batch Loss:     2.066353 Tokens per Sec:     7595, Lr: 0.000300\n",
      "2020-02-20 14:30:32,291 Epoch  16 Step:    23600 Batch Loss:     1.938859 Tokens per Sec:     7612, Lr: 0.000300\n",
      "2020-02-20 14:31:00,714 Epoch  16 Step:    23700 Batch Loss:     1.713047 Tokens per Sec:     7437, Lr: 0.000300\n",
      "2020-02-20 14:31:29,704 Epoch  16 Step:    23800 Batch Loss:     1.987713 Tokens per Sec:     7574, Lr: 0.000300\n",
      "2020-02-20 14:31:58,520 Epoch  16 Step:    23900 Batch Loss:     1.617335 Tokens per Sec:     7662, Lr: 0.000300\n",
      "2020-02-20 14:32:27,342 Epoch  16 Step:    24000 Batch Loss:     2.072048 Tokens per Sec:     7589, Lr: 0.000300\n",
      "2020-02-20 14:33:55,514 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:33:55,514 Saving new checkpoint.\n",
      "2020-02-20 14:33:55,735 Example #0\n",
      "2020-02-20 14:33:55,735 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:33:55,735 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:33:55,735 \tHypothesis: Luwo chikene ne Jehova dwaro ni onyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 14:33:55,736 Example #1\n",
      "2020-02-20 14:33:55,736 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:33:55,736 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:33:55,736 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuonwa kod winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 14:33:55,736 Example #2\n",
      "2020-02-20 14:33:55,736 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:33:55,736 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:33:55,736 \tHypothesis: Ng’eyo ni Muma wacho adier nyalo bedo gi teko mar ng’eyo gik manie chuny .\n",
      "2020-02-20 14:33:55,736 Example #3\n",
      "2020-02-20 14:33:55,737 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:33:55,737 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:33:55,737 \tHypothesis: Nondike e wi “ weche duto mag Nyasaye ma ne wuoyo kuom jonabi maler , ” ma ne wuoyo kuom jonabi maler .\n",
      "2020-02-20 14:33:55,737 Validation result at epoch  16, step    24000: bleu:  20.11, loss: 45200.4805, ppl:   6.2579, duration: 88.3943s\n",
      "2020-02-20 14:34:24,395 Epoch  16 Step:    24100 Batch Loss:     2.104906 Tokens per Sec:     7481, Lr: 0.000300\n",
      "2020-02-20 14:34:53,164 Epoch  16 Step:    24200 Batch Loss:     1.972242 Tokens per Sec:     7663, Lr: 0.000300\n",
      "2020-02-20 14:35:21,945 Epoch  16 Step:    24300 Batch Loss:     2.164508 Tokens per Sec:     7640, Lr: 0.000300\n",
      "2020-02-20 14:35:28,695 Epoch  16: total training loss 2886.69\n",
      "2020-02-20 14:35:28,695 EPOCH 17\n",
      "2020-02-20 14:35:51,053 Epoch  17 Step:    24400 Batch Loss:     1.791167 Tokens per Sec:     7492, Lr: 0.000300\n",
      "2020-02-20 14:36:19,860 Epoch  17 Step:    24500 Batch Loss:     1.838502 Tokens per Sec:     7565, Lr: 0.000300\n",
      "2020-02-20 14:36:48,940 Epoch  17 Step:    24600 Batch Loss:     1.565369 Tokens per Sec:     7661, Lr: 0.000300\n",
      "2020-02-20 14:37:17,795 Epoch  17 Step:    24700 Batch Loss:     1.729771 Tokens per Sec:     7598, Lr: 0.000300\n",
      "2020-02-20 14:37:46,643 Epoch  17 Step:    24800 Batch Loss:     2.019106 Tokens per Sec:     7539, Lr: 0.000300\n",
      "2020-02-20 14:38:15,421 Epoch  17 Step:    24900 Batch Loss:     2.044064 Tokens per Sec:     7574, Lr: 0.000300\n",
      "2020-02-20 14:38:44,177 Epoch  17 Step:    25000 Batch Loss:     2.007406 Tokens per Sec:     7568, Lr: 0.000300\n",
      "2020-02-20 14:40:12,390 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:40:12,390 Saving new checkpoint.\n",
      "2020-02-20 14:40:12,612 Example #0\n",
      "2020-02-20 14:40:12,612 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:40:12,612 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:40:12,612 \tHypothesis: Luwo chikene ne gik ma Jehova dwaro ni mondo onyis ni chike Nyasaye ema owinjore .\n",
      "2020-02-20 14:40:12,613 Example #1\n",
      "2020-02-20 14:40:12,613 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:40:12,613 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:40:12,613 \tHypothesis: Bende , amiyo luor kaka wuonwa kod winjruok manie kindwa kode . ”\n",
      "2020-02-20 14:40:12,613 Example #2\n",
      "2020-02-20 14:40:12,613 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:40:12,613 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:40:12,613 \tHypothesis: Ng’eyo ni Muma wacho adier nyalo bedo gima duong ’ ahinya e fwenyo gik manie chunywa .\n",
      "2020-02-20 14:40:12,613 Example #3\n",
      "2020-02-20 14:40:12,614 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:40:12,614 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:40:12,614 \tHypothesis: Noyudo osendiko weche ma Nyasaye nowacho e wi “ weche duto ma Nyasaye nowuoyo kuom janabi maler , ma ne en janabi maler . ”\n",
      "2020-02-20 14:40:12,614 Validation result at epoch  17, step    25000: bleu:  19.91, loss: 44906.8984, ppl:   6.1838, duration: 88.4365s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 14:40:41,844 Epoch  17 Step:    25100 Batch Loss:     1.930991 Tokens per Sec:     7626, Lr: 0.000300\n",
      "2020-02-20 14:41:10,728 Epoch  17 Step:    25200 Batch Loss:     1.790509 Tokens per Sec:     7552, Lr: 0.000300\n",
      "2020-02-20 14:41:39,542 Epoch  17 Step:    25300 Batch Loss:     1.659317 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 14:42:08,450 Epoch  17 Step:    25400 Batch Loss:     1.727902 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 14:42:37,179 Epoch  17 Step:    25500 Batch Loss:     2.255066 Tokens per Sec:     7499, Lr: 0.000300\n",
      "2020-02-20 14:43:06,080 Epoch  17 Step:    25600 Batch Loss:     1.129977 Tokens per Sec:     7543, Lr: 0.000300\n",
      "2020-02-20 14:43:35,073 Epoch  17 Step:    25700 Batch Loss:     1.938497 Tokens per Sec:     7640, Lr: 0.000300\n",
      "2020-02-20 14:44:04,141 Epoch  17 Step:    25800 Batch Loss:     1.842105 Tokens per Sec:     7640, Lr: 0.000300\n",
      "2020-02-20 14:44:15,739 Epoch  17: total training loss 2840.30\n",
      "2020-02-20 14:44:15,740 EPOCH 18\n",
      "2020-02-20 14:44:32,707 Epoch  18 Step:    25900 Batch Loss:     1.698621 Tokens per Sec:     7253, Lr: 0.000300\n",
      "2020-02-20 14:45:01,407 Epoch  18 Step:    26000 Batch Loss:     1.913930 Tokens per Sec:     7689, Lr: 0.000300\n",
      "2020-02-20 14:46:29,459 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:46:29,459 Saving new checkpoint.\n",
      "2020-02-20 14:46:29,674 Example #0\n",
      "2020-02-20 14:46:29,675 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:46:29,675 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:46:29,675 \tHypothesis: Chiwruok ma Jehova dwaro ni mondo okonygi ng’eyo ni chike Nyasaye ema owinjore .\n",
      "2020-02-20 14:46:29,675 Example #1\n",
      "2020-02-20 14:46:29,675 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:46:29,675 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:46:29,675 \tHypothesis: Bende , omiya thuolo mar miyo luor kaka wuonwa kod winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 14:46:29,676 Example #2\n",
      "2020-02-20 14:46:29,676 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:46:29,676 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:46:29,676 \tHypothesis: Ng’eyo ni Muma wacho adier nyalo konyowa ng’eyo kaka chuny ng’ato nyalo bedo gi teko .\n",
      "2020-02-20 14:46:29,676 Example #3\n",
      "2020-02-20 14:46:29,676 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:46:29,676 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:46:29,676 \tHypothesis: Nondike e wi “ weche duto mag Nyasaye ma Nyasaye nowuoyo kuome e dhoge machon . ”\n",
      "2020-02-20 14:46:29,676 Validation result at epoch  18, step    26000: bleu:  19.35, loss: 44817.0664, ppl:   6.1613, duration: 88.2689s\n",
      "2020-02-20 14:46:58,596 Epoch  18 Step:    26100 Batch Loss:     2.105821 Tokens per Sec:     7544, Lr: 0.000300\n",
      "2020-02-20 14:47:27,182 Epoch  18 Step:    26200 Batch Loss:     2.101782 Tokens per Sec:     7616, Lr: 0.000300\n",
      "2020-02-20 14:47:55,984 Epoch  18 Step:    26300 Batch Loss:     1.876869 Tokens per Sec:     7510, Lr: 0.000300\n",
      "2020-02-20 14:48:24,363 Epoch  18 Step:    26400 Batch Loss:     1.818757 Tokens per Sec:     7522, Lr: 0.000300\n",
      "2020-02-20 14:48:53,156 Epoch  18 Step:    26500 Batch Loss:     2.156519 Tokens per Sec:     7487, Lr: 0.000300\n",
      "2020-02-20 14:49:21,762 Epoch  18 Step:    26600 Batch Loss:     1.930380 Tokens per Sec:     7538, Lr: 0.000300\n",
      "2020-02-20 14:49:50,597 Epoch  18 Step:    26700 Batch Loss:     1.636875 Tokens per Sec:     7634, Lr: 0.000300\n",
      "2020-02-20 14:50:19,454 Epoch  18 Step:    26800 Batch Loss:     2.019928 Tokens per Sec:     7613, Lr: 0.000300\n",
      "2020-02-20 14:50:48,100 Epoch  18 Step:    26900 Batch Loss:     2.064134 Tokens per Sec:     7606, Lr: 0.000300\n",
      "2020-02-20 14:51:17,265 Epoch  18 Step:    27000 Batch Loss:     2.183713 Tokens per Sec:     7635, Lr: 0.000300\n",
      "2020-02-20 14:52:45,420 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:52:45,420 Saving new checkpoint.\n",
      "2020-02-20 14:52:45,638 Example #0\n",
      "2020-02-20 14:52:45,638 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:52:45,639 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:52:45,639 \tHypothesis: Luwo chikene , dwaro ni Jehova ochik gik ma odwaro , en ni chike mag Nyasaye ema owinjore .\n",
      "2020-02-20 14:52:45,639 Example #1\n",
      "2020-02-20 14:52:45,639 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:52:45,639 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:52:45,639 \tHypothesis: Bende , en gima miyo amiye luor kaka wuonwa koda wiye . ”\n",
      "2020-02-20 14:52:45,639 Example #2\n",
      "2020-02-20 14:52:45,639 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:52:45,639 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:52:45,639 \tHypothesis: Ng’eyo ni wan wawegi e ler ma Ndiko wacho nyalo bedo gi teko mar fwenyo chuny mar chuny .\n",
      "2020-02-20 14:52:45,640 Example #3\n",
      "2020-02-20 14:52:45,640 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:52:45,640 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:52:45,640 \tHypothesis: Ne ondiko e wi “ gik moko duto ma Nyasaye nowuoyo kuome kokalo kuom jonabi maler , ma ne en janabi maler . ”\n",
      "2020-02-20 14:52:45,640 Validation result at epoch  18, step    27000: bleu:  20.44, loss: 44218.1758, ppl:   6.0134, duration: 88.3739s\n",
      "2020-02-20 14:53:14,199 Epoch  18 Step:    27100 Batch Loss:     1.837035 Tokens per Sec:     7515, Lr: 0.000300\n",
      "2020-02-20 14:53:43,290 Epoch  18 Step:    27200 Batch Loss:     1.722504 Tokens per Sec:     7614, Lr: 0.000300\n",
      "2020-02-20 14:54:12,258 Epoch  18 Step:    27300 Batch Loss:     1.830405 Tokens per Sec:     7600, Lr: 0.000300\n",
      "2020-02-20 14:54:30,877 Epoch  18: total training loss 2821.93\n",
      "2020-02-20 14:54:30,877 EPOCH 19\n",
      "2020-02-20 14:54:41,573 Epoch  19 Step:    27400 Batch Loss:     1.818916 Tokens per Sec:     7436, Lr: 0.000300\n",
      "2020-02-20 14:55:10,496 Epoch  19 Step:    27500 Batch Loss:     1.644370 Tokens per Sec:     7690, Lr: 0.000300\n",
      "2020-02-20 14:55:39,376 Epoch  19 Step:    27600 Batch Loss:     1.980896 Tokens per Sec:     7614, Lr: 0.000300\n",
      "2020-02-20 14:56:07,923 Epoch  19 Step:    27700 Batch Loss:     1.704031 Tokens per Sec:     7469, Lr: 0.000300\n",
      "2020-02-20 14:56:36,653 Epoch  19 Step:    27800 Batch Loss:     1.831275 Tokens per Sec:     7393, Lr: 0.000300\n",
      "2020-02-20 14:57:05,844 Epoch  19 Step:    27900 Batch Loss:     2.040896 Tokens per Sec:     7743, Lr: 0.000300\n",
      "2020-02-20 14:57:34,599 Epoch  19 Step:    28000 Batch Loss:     1.963816 Tokens per Sec:     7609, Lr: 0.000300\n",
      "2020-02-20 14:59:02,671 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 14:59:02,671 Saving new checkpoint.\n",
      "2020-02-20 14:59:02,895 Example #0\n",
      "2020-02-20 14:59:02,896 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 14:59:02,896 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 14:59:02,896 \tHypothesis: Luwo chike Jehova e yo makare , onego onyis ni chike Nyasaye nigi paro mowinjore .\n",
      "2020-02-20 14:59:02,896 Example #1\n",
      "2020-02-20 14:59:02,896 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 14:59:02,896 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 14:59:02,896 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuoro kod wiye . ”\n",
      "2020-02-20 14:59:02,896 Example #2\n",
      "2020-02-20 14:59:02,897 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 14:59:02,897 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 14:59:02,897 \tHypothesis: Nono adierago ei Ndiko ma wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 14:59:02,897 Example #3\n",
      "2020-02-20 14:59:02,897 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 14:59:02,897 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 14:59:02,897 \tHypothesis: Nondike e wi “ gik moko duto ma Nyasaye nowuoyo kuome e dhoge machon . ”\n",
      "2020-02-20 14:59:02,897 Validation result at epoch  19, step    28000: bleu:  20.51, loss: 44157.7266, ppl:   5.9986, duration: 88.2978s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 14:59:31,948 Epoch  19 Step:    28100 Batch Loss:     1.910940 Tokens per Sec:     7711, Lr: 0.000300\n",
      "2020-02-20 15:00:00,620 Epoch  19 Step:    28200 Batch Loss:     1.857561 Tokens per Sec:     7624, Lr: 0.000300\n",
      "2020-02-20 15:00:29,051 Epoch  19 Step:    28300 Batch Loss:     1.970976 Tokens per Sec:     7407, Lr: 0.000300\n",
      "2020-02-20 15:00:58,128 Epoch  19 Step:    28400 Batch Loss:     1.691468 Tokens per Sec:     7570, Lr: 0.000300\n",
      "2020-02-20 15:01:26,796 Epoch  19 Step:    28500 Batch Loss:     1.886670 Tokens per Sec:     7515, Lr: 0.000300\n",
      "2020-02-20 15:01:55,426 Epoch  19 Step:    28600 Batch Loss:     2.225640 Tokens per Sec:     7532, Lr: 0.000300\n",
      "2020-02-20 15:02:24,377 Epoch  19 Step:    28700 Batch Loss:     1.406204 Tokens per Sec:     7593, Lr: 0.000300\n",
      "2020-02-20 15:02:53,014 Epoch  19 Step:    28800 Batch Loss:     1.695791 Tokens per Sec:     7448, Lr: 0.000300\n",
      "2020-02-20 15:03:18,084 Epoch  19: total training loss 2790.81\n",
      "2020-02-20 15:03:18,085 EPOCH 20\n",
      "2020-02-20 15:03:22,124 Epoch  20 Step:    28900 Batch Loss:     1.790626 Tokens per Sec:     7397, Lr: 0.000300\n",
      "2020-02-20 15:03:50,648 Epoch  20 Step:    29000 Batch Loss:     1.866627 Tokens per Sec:     7535, Lr: 0.000300\n",
      "2020-02-20 15:05:18,692 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:05:18,692 Saving new checkpoint.\n",
      "2020-02-20 15:05:18,911 Example #0\n",
      "2020-02-20 15:05:18,911 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:05:18,911 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:05:18,911 \tHypothesis: Luwo chik ma Jehova dwaro ni mondo onyis ni chike mag Nyasaye owinjore .\n",
      "2020-02-20 15:05:18,911 Example #1\n",
      "2020-02-20 15:05:18,912 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:05:18,912 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:05:18,912 \tHypothesis: Bende , omiya thuolo mar miyo amiye luor kaka wuonwa koda wiye . ”\n",
      "2020-02-20 15:05:18,912 Example #2\n",
      "2020-02-20 15:05:18,912 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:05:18,912 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:05:18,912 \tHypothesis: Ng’eyo gik ma Ndiko wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 15:05:18,912 Example #3\n",
      "2020-02-20 15:05:18,912 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:05:18,913 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:05:18,913 \tHypothesis: Ne ondiko e wi “ weche duto mag Nyasaye ma ne wuoyo kuome e dho janabi maler , ma ne en janabi maler . ”\n",
      "2020-02-20 15:05:18,913 Validation result at epoch  20, step    29000: bleu:  20.97, loss: 43801.0352, ppl:   5.9125, duration: 88.2640s\n",
      "2020-02-20 15:05:47,655 Epoch  20 Step:    29100 Batch Loss:     1.752785 Tokens per Sec:     7613, Lr: 0.000300\n",
      "2020-02-20 15:06:16,751 Epoch  20 Step:    29200 Batch Loss:     1.508315 Tokens per Sec:     7679, Lr: 0.000300\n",
      "2020-02-20 15:06:45,926 Epoch  20 Step:    29300 Batch Loss:     1.816944 Tokens per Sec:     7820, Lr: 0.000300\n",
      "2020-02-20 15:07:14,827 Epoch  20 Step:    29400 Batch Loss:     1.822889 Tokens per Sec:     7560, Lr: 0.000300\n",
      "2020-02-20 15:07:43,516 Epoch  20 Step:    29500 Batch Loss:     2.019667 Tokens per Sec:     7581, Lr: 0.000300\n",
      "2020-02-20 15:08:12,460 Epoch  20 Step:    29600 Batch Loss:     1.640045 Tokens per Sec:     7638, Lr: 0.000300\n",
      "2020-02-20 15:08:40,786 Epoch  20 Step:    29700 Batch Loss:     2.031919 Tokens per Sec:     7375, Lr: 0.000300\n",
      "2020-02-20 15:09:09,955 Epoch  20 Step:    29800 Batch Loss:     1.888292 Tokens per Sec:     7733, Lr: 0.000300\n",
      "2020-02-20 15:09:38,439 Epoch  20 Step:    29900 Batch Loss:     1.080145 Tokens per Sec:     7422, Lr: 0.000300\n",
      "2020-02-20 15:10:07,055 Epoch  20 Step:    30000 Batch Loss:     2.215950 Tokens per Sec:     7469, Lr: 0.000300\n",
      "2020-02-20 15:11:35,124 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:11:35,124 Saving new checkpoint.\n",
      "2020-02-20 15:11:35,345 Example #0\n",
      "2020-02-20 15:11:35,345 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:11:35,345 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:11:35,345 \tHypothesis: Luwo chike Jehova e yo ma dwaro ni wanyis ni chike Nyasaye ok owinjore .\n",
      "2020-02-20 15:11:35,346 Example #1\n",
      "2020-02-20 15:11:35,346 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:11:35,346 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:11:35,346 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuonwa koda wiye . ”\n",
      "2020-02-20 15:11:35,346 Example #2\n",
      "2020-02-20 15:11:35,347 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:11:35,347 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:11:35,347 \tHypothesis: Ka wafwenyo ni wan wawegi warito ler mar Ndiko , mano nyalo miyo wafweny matut gik manie chunywa .\n",
      "2020-02-20 15:11:35,347 Example #3\n",
      "2020-02-20 15:11:35,347 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:11:35,347 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:11:35,347 \tHypothesis: Noyudo osendiko e wi “ weche duto mag Nyasaye ma Nyasaye nowuoyo kuome kokalo kuom jonabi maler . ”\n",
      "2020-02-20 15:11:35,348 Validation result at epoch  20, step    30000: bleu:  21.01, loss: 43687.1055, ppl:   5.8852, duration: 88.2914s\n",
      "2020-02-20 15:12:03,930 Epoch  20 Step:    30100 Batch Loss:     1.640167 Tokens per Sec:     7534, Lr: 0.000300\n",
      "2020-02-20 15:12:32,293 Epoch  20 Step:    30200 Batch Loss:     1.806264 Tokens per Sec:     7464, Lr: 0.000300\n",
      "2020-02-20 15:13:01,063 Epoch  20 Step:    30300 Batch Loss:     1.790595 Tokens per Sec:     7671, Lr: 0.000300\n",
      "2020-02-20 15:13:29,516 Epoch  20 Step:    30400 Batch Loss:     1.406210 Tokens per Sec:     7558, Lr: 0.000300\n",
      "2020-02-20 15:13:32,776 Epoch  20: total training loss 2756.82\n",
      "2020-02-20 15:13:32,776 EPOCH 21\n",
      "2020-02-20 15:13:58,627 Epoch  21 Step:    30500 Batch Loss:     1.913261 Tokens per Sec:     7420, Lr: 0.000300\n",
      "2020-02-20 15:14:27,334 Epoch  21 Step:    30600 Batch Loss:     1.737304 Tokens per Sec:     7541, Lr: 0.000300\n",
      "2020-02-20 15:14:56,442 Epoch  21 Step:    30700 Batch Loss:     1.684234 Tokens per Sec:     7634, Lr: 0.000300\n",
      "2020-02-20 15:15:25,375 Epoch  21 Step:    30800 Batch Loss:     2.049634 Tokens per Sec:     7664, Lr: 0.000300\n",
      "2020-02-20 15:15:54,106 Epoch  21 Step:    30900 Batch Loss:     1.679987 Tokens per Sec:     7571, Lr: 0.000300\n",
      "2020-02-20 15:16:22,846 Epoch  21 Step:    31000 Batch Loss:     1.611002 Tokens per Sec:     7605, Lr: 0.000300\n",
      "2020-02-20 15:17:50,653 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:17:50,654 Saving new checkpoint.\n",
      "2020-02-20 15:17:50,869 Example #0\n",
      "2020-02-20 15:17:50,870 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:17:50,870 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:17:50,870 \tHypothesis: Luwo chike Jehova e yo ma Jehova dwaro ni mondo okonywa bedo gi paro mowinjore .\n",
      "2020-02-20 15:17:50,870 Example #1\n",
      "2020-02-20 15:17:50,870 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:17:50,870 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:17:50,870 \tHypothesis: Bende , omiya rieko mar miye luor kaka wuonwa gi wiye . ”\n",
      "2020-02-20 15:17:50,871 Example #2\n",
      "2020-02-20 15:17:50,871 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:17:50,871 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:17:50,871 \tHypothesis: Ka wafwenyo ler mar weche manie Ndiko , wanyalo fwenyo ni chunywa nigi teko ahinya .\n",
      "2020-02-20 15:17:50,871 Example #3\n",
      "2020-02-20 15:17:50,871 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:17:50,871 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:17:50,871 \tHypothesis: Noyudo osendiko e wi “ weche duto mag Nyasaye ma Nyasaye nowuoyo kuome e dho jonabi maler kuom kinde machon . ”\n",
      "2020-02-20 15:17:50,871 Validation result at epoch  21, step    31000: bleu:  20.83, loss: 43309.3750, ppl:   5.7957, duration: 88.0246s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 15:18:19,622 Epoch  21 Step:    31100 Batch Loss:     1.819182 Tokens per Sec:     7524, Lr: 0.000300\n",
      "2020-02-20 15:18:48,484 Epoch  21 Step:    31200 Batch Loss:     1.671183 Tokens per Sec:     7697, Lr: 0.000300\n",
      "2020-02-20 15:19:17,193 Epoch  21 Step:    31300 Batch Loss:     1.073250 Tokens per Sec:     7488, Lr: 0.000300\n",
      "2020-02-20 15:19:45,994 Epoch  21 Step:    31400 Batch Loss:     1.645528 Tokens per Sec:     7604, Lr: 0.000300\n",
      "2020-02-20 15:20:14,749 Epoch  21 Step:    31500 Batch Loss:     1.817031 Tokens per Sec:     7573, Lr: 0.000300\n",
      "2020-02-20 15:20:43,542 Epoch  21 Step:    31600 Batch Loss:     1.894629 Tokens per Sec:     7575, Lr: 0.000300\n",
      "2020-02-20 15:21:12,418 Epoch  21 Step:    31700 Batch Loss:     1.885237 Tokens per Sec:     7576, Lr: 0.000300\n",
      "2020-02-20 15:21:41,376 Epoch  21 Step:    31800 Batch Loss:     1.353623 Tokens per Sec:     7636, Lr: 0.000300\n",
      "2020-02-20 15:22:10,286 Epoch  21 Step:    31900 Batch Loss:     1.591935 Tokens per Sec:     7582, Lr: 0.000300\n",
      "2020-02-20 15:22:18,431 Epoch  21: total training loss 2717.05\n",
      "2020-02-20 15:22:18,431 EPOCH 22\n",
      "2020-02-20 15:22:39,288 Epoch  22 Step:    32000 Batch Loss:     1.760571 Tokens per Sec:     7427, Lr: 0.000300\n",
      "2020-02-20 15:24:07,383 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:24:07,384 Saving new checkpoint.\n",
      "2020-02-20 15:24:07,606 Example #0\n",
      "2020-02-20 15:24:07,606 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:24:07,606 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:24:07,606 \tHypothesis: Luwo chikene makare , onego onyis ni chike Nyasaye ok nyal winjo tiende .\n",
      "2020-02-20 15:24:07,606 Example #1\n",
      "2020-02-20 15:24:07,607 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:24:07,607 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:24:07,607 \tHypothesis: Bende , omiya thuolo momedore mar miyo Nyasaye luor kendo tayo winjruokna kode . ”\n",
      "2020-02-20 15:24:07,607 Example #2\n",
      "2020-02-20 15:24:07,607 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:24:07,607 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:24:07,607 \tHypothesis: Ng’eyo ni Muma wacho adier nyalo bedo gi teko mar ng’eyo chuny ng’ato .\n",
      "2020-02-20 15:24:07,607 Example #3\n",
      "2020-02-20 15:24:07,607 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:24:07,607 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:24:07,608 \tHypothesis: Noyudo osendiko e wi “ gik moko duto ma Nyasaye nowuoyo kuome kokalo kuom jonabi maler . ”\n",
      "2020-02-20 15:24:07,608 Validation result at epoch  22, step    32000: bleu:  21.35, loss: 43100.0508, ppl:   5.7467, duration: 88.3187s\n",
      "2020-02-20 15:24:36,222 Epoch  22 Step:    32100 Batch Loss:     1.414382 Tokens per Sec:     7497, Lr: 0.000300\n",
      "2020-02-20 15:25:04,712 Epoch  22 Step:    32200 Batch Loss:     1.545235 Tokens per Sec:     7444, Lr: 0.000300\n",
      "2020-02-20 15:25:33,666 Epoch  22 Step:    32300 Batch Loss:     2.022698 Tokens per Sec:     7642, Lr: 0.000300\n",
      "2020-02-20 15:26:02,242 Epoch  22 Step:    32400 Batch Loss:     2.208554 Tokens per Sec:     7568, Lr: 0.000300\n",
      "2020-02-20 15:26:31,015 Epoch  22 Step:    32500 Batch Loss:     1.741829 Tokens per Sec:     7576, Lr: 0.000300\n",
      "2020-02-20 15:26:59,857 Epoch  22 Step:    32600 Batch Loss:     1.308279 Tokens per Sec:     7590, Lr: 0.000300\n",
      "2020-02-20 15:27:28,962 Epoch  22 Step:    32700 Batch Loss:     1.890848 Tokens per Sec:     7652, Lr: 0.000300\n",
      "2020-02-20 15:27:57,781 Epoch  22 Step:    32800 Batch Loss:     1.702807 Tokens per Sec:     7751, Lr: 0.000300\n",
      "2020-02-20 15:28:26,449 Epoch  22 Step:    32900 Batch Loss:     2.175448 Tokens per Sec:     7541, Lr: 0.000300\n",
      "2020-02-20 15:28:55,299 Epoch  22 Step:    33000 Batch Loss:     1.553985 Tokens per Sec:     7687, Lr: 0.000300\n",
      "2020-02-20 15:30:23,371 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:30:23,371 Saving new checkpoint.\n",
      "2020-02-20 15:30:23,589 Example #0\n",
      "2020-02-20 15:30:23,589 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:30:23,589 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:30:23,589 \tHypothesis: Luwo chike Jehova en gima dwaro ni mondo giluw chikene .\n",
      "2020-02-20 15:30:23,590 Example #1\n",
      "2020-02-20 15:30:23,590 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:30:23,590 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:30:23,590 \tHypothesis: Bende , omiya thuolo momedore mar miye luor kaka wuonwa gi wiye . ”\n",
      "2020-02-20 15:30:23,590 Example #2\n",
      "2020-02-20 15:30:23,590 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:30:23,590 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:30:23,590 \tHypothesis: Nono gik ma Ndiko wacho nyalo konyo ahinya e fwenyo gik ma chunywa dwaro .\n",
      "2020-02-20 15:30:23,590 Example #3\n",
      "2020-02-20 15:30:23,591 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:30:23,591 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:30:23,591 \tHypothesis: Ne ondiko e wi “ gik moko duto ma Nyasaye nowuoyo kuome kokalo kuom dho jonabi maler , ma ne en jonabi machon . ”\n",
      "2020-02-20 15:30:23,591 Validation result at epoch  22, step    33000: bleu:  20.99, loss: 42939.6875, ppl:   5.7094, duration: 88.2914s\n",
      "2020-02-20 15:30:52,353 Epoch  22 Step:    33100 Batch Loss:     2.253767 Tokens per Sec:     7501, Lr: 0.000300\n",
      "2020-02-20 15:31:21,146 Epoch  22 Step:    33200 Batch Loss:     2.039063 Tokens per Sec:     7529, Lr: 0.000300\n",
      "2020-02-20 15:31:49,986 Epoch  22 Step:    33300 Batch Loss:     1.518212 Tokens per Sec:     7588, Lr: 0.000300\n",
      "2020-02-20 15:32:18,995 Epoch  22 Step:    33400 Batch Loss:     1.767002 Tokens per Sec:     7592, Lr: 0.000300\n",
      "2020-02-20 15:32:33,246 Epoch  22: total training loss 2699.78\n",
      "2020-02-20 15:32:33,246 EPOCH 23\n",
      "2020-02-20 15:32:47,958 Epoch  23 Step:    33500 Batch Loss:     1.491392 Tokens per Sec:     7456, Lr: 0.000300\n",
      "2020-02-20 15:33:16,992 Epoch  23 Step:    33600 Batch Loss:     1.746894 Tokens per Sec:     7591, Lr: 0.000300\n",
      "2020-02-20 15:33:45,691 Epoch  23 Step:    33700 Batch Loss:     1.184824 Tokens per Sec:     7602, Lr: 0.000300\n",
      "2020-02-20 15:34:14,468 Epoch  23 Step:    33800 Batch Loss:     1.476453 Tokens per Sec:     7481, Lr: 0.000300\n",
      "2020-02-20 15:34:42,872 Epoch  23 Step:    33900 Batch Loss:     1.657290 Tokens per Sec:     7496, Lr: 0.000300\n",
      "2020-02-20 15:35:11,671 Epoch  23 Step:    34000 Batch Loss:     1.806342 Tokens per Sec:     7676, Lr: 0.000300\n",
      "2020-02-20 15:36:39,685 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:36:39,685 Saving new checkpoint.\n",
      "2020-02-20 15:36:39,899 Example #0\n",
      "2020-02-20 15:36:39,899 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:36:39,899 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:36:39,900 \tHypothesis: Luwo chike Jehova dwaro ni ginyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 15:36:39,900 Example #1\n",
      "2020-02-20 15:36:39,900 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:36:39,900 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:36:39,900 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuonwa koda wiye . ”\n",
      "2020-02-20 15:36:39,900 Example #2\n",
      "2020-02-20 15:36:39,900 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:36:39,900 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:36:39,900 \tHypothesis: Nono gik ma Ndiko wacho nyalo konyowa ng’eyo kaka chuny ng’ato nyalo bedo gi teko .\n",
      "2020-02-20 15:36:39,900 Example #3\n",
      "2020-02-20 15:36:39,901 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:36:39,901 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:36:39,901 \tHypothesis: Noyudo osendiko e wi “ gik moko duto ma Nyasaye nowuoyo kuom jonabi maler kuom kinde machon . ”\n",
      "2020-02-20 15:36:39,901 Validation result at epoch  23, step    34000: bleu:  21.52, loss: 42751.1289, ppl:   5.6659, duration: 88.2290s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 15:37:08,695 Epoch  23 Step:    34100 Batch Loss:     1.914464 Tokens per Sec:     7584, Lr: 0.000300\n",
      "2020-02-20 15:37:37,404 Epoch  23 Step:    34200 Batch Loss:     1.760642 Tokens per Sec:     7584, Lr: 0.000300\n",
      "2020-02-20 15:38:06,019 Epoch  23 Step:    34300 Batch Loss:     1.812148 Tokens per Sec:     7586, Lr: 0.000300\n",
      "2020-02-20 15:38:34,760 Epoch  23 Step:    34400 Batch Loss:     1.241874 Tokens per Sec:     7601, Lr: 0.000300\n",
      "2020-02-20 15:39:03,539 Epoch  23 Step:    34500 Batch Loss:     1.844906 Tokens per Sec:     7624, Lr: 0.000300\n",
      "2020-02-20 15:39:32,884 Epoch  23 Step:    34600 Batch Loss:     1.787987 Tokens per Sec:     7863, Lr: 0.000300\n",
      "2020-02-20 15:40:01,826 Epoch  23 Step:    34700 Batch Loss:     1.775381 Tokens per Sec:     7588, Lr: 0.000300\n",
      "2020-02-20 15:40:30,810 Epoch  23 Step:    34800 Batch Loss:     1.828186 Tokens per Sec:     7622, Lr: 0.000300\n",
      "2020-02-20 15:40:59,724 Epoch  23 Step:    34900 Batch Loss:     1.916007 Tokens per Sec:     7516, Lr: 0.000300\n",
      "2020-02-20 15:41:19,055 Epoch  23: total training loss 2670.62\n",
      "2020-02-20 15:41:19,055 EPOCH 24\n",
      "2020-02-20 15:41:28,576 Epoch  24 Step:    35000 Batch Loss:     1.632923 Tokens per Sec:     7289, Lr: 0.000300\n",
      "2020-02-20 15:42:56,702 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:42:56,702 Saving new checkpoint.\n",
      "2020-02-20 15:42:56,920 Example #0\n",
      "2020-02-20 15:42:56,920 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:42:56,920 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:42:56,920 \tHypothesis: Luwo chikene , dwaro ni wanyis ni chike Nyasaye nigi paro mowinjore .\n",
      "2020-02-20 15:42:56,921 Example #1\n",
      "2020-02-20 15:42:56,921 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:42:56,921 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:42:56,921 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuonwa koda winjruok maber gi Nyasaye . ”\n",
      "2020-02-20 15:42:56,921 Example #2\n",
      "2020-02-20 15:42:56,921 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:42:56,921 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:42:56,921 \tHypothesis: Nono gik ma Ndiko wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 15:42:56,921 Example #3\n",
      "2020-02-20 15:42:56,922 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:42:56,922 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:42:56,922 \tHypothesis: Noyudo osendiko e wi “ gik moko duto ma Nyasaye nowuoyo kuome e dho jonabi maler . ”\n",
      "2020-02-20 15:42:56,922 Validation result at epoch  24, step    35000: bleu:  21.60, loss: 42456.4180, ppl:   5.5986, duration: 88.3449s\n",
      "2020-02-20 15:43:25,825 Epoch  24 Step:    35100 Batch Loss:     2.230939 Tokens per Sec:     7603, Lr: 0.000300\n",
      "2020-02-20 15:43:54,911 Epoch  24 Step:    35200 Batch Loss:     1.771439 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 15:44:23,615 Epoch  24 Step:    35300 Batch Loss:     2.057425 Tokens per Sec:     7470, Lr: 0.000300\n",
      "2020-02-20 15:44:52,359 Epoch  24 Step:    35400 Batch Loss:     2.026919 Tokens per Sec:     7631, Lr: 0.000300\n",
      "2020-02-20 15:45:20,898 Epoch  24 Step:    35500 Batch Loss:     1.822903 Tokens per Sec:     7536, Lr: 0.000300\n",
      "2020-02-20 15:45:49,847 Epoch  24 Step:    35600 Batch Loss:     1.721725 Tokens per Sec:     7618, Lr: 0.000300\n",
      "2020-02-20 15:46:18,354 Epoch  24 Step:    35700 Batch Loss:     1.844272 Tokens per Sec:     7455, Lr: 0.000300\n",
      "2020-02-20 15:46:46,754 Epoch  24 Step:    35800 Batch Loss:     1.998473 Tokens per Sec:     7488, Lr: 0.000300\n",
      "2020-02-20 15:47:16,004 Epoch  24 Step:    35900 Batch Loss:     1.597854 Tokens per Sec:     7708, Lr: 0.000300\n",
      "2020-02-20 15:47:44,855 Epoch  24 Step:    36000 Batch Loss:     1.540648 Tokens per Sec:     7538, Lr: 0.000300\n",
      "2020-02-20 15:49:13,031 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:49:13,031 Saving new checkpoint.\n",
      "2020-02-20 15:49:13,250 Example #0\n",
      "2020-02-20 15:49:13,250 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:49:13,250 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:49:13,250 \tHypothesis: Luwo chikene , onego onyis ni chike Nyasaye ema owinjore .\n",
      "2020-02-20 15:49:13,250 Example #1\n",
      "2020-02-20 15:49:13,250 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:49:13,250 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:49:13,250 \tHypothesis: Bende , omiya thuolo mar miye luor kaka wuonwa kod wiye . ”\n",
      "2020-02-20 15:49:13,251 Example #2\n",
      "2020-02-20 15:49:13,251 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:49:13,251 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:49:13,251 \tHypothesis: Neno gik ma Ndiko wacho nyalo konyowa ng’eyo kaka chunywa chalo .\n",
      "2020-02-20 15:49:13,251 Example #3\n",
      "2020-02-20 15:49:13,251 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 15:49:13,251 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:49:13,251 \tHypothesis: Noyudo osendiko kuom “ weche duto ma Nyasaye nowuoyo kuome kokalo kuom jonabi maler ma ne en jonabi maler . ”\n",
      "2020-02-20 15:49:13,251 Validation result at epoch  24, step    36000: bleu:  21.19, loss: 42349.5820, ppl:   5.5743, duration: 88.3960s\n",
      "2020-02-20 15:49:42,256 Epoch  24 Step:    36100 Batch Loss:     1.857429 Tokens per Sec:     7654, Lr: 0.000300\n",
      "2020-02-20 15:50:11,323 Epoch  24 Step:    36200 Batch Loss:     1.692087 Tokens per Sec:     7579, Lr: 0.000300\n",
      "2020-02-20 15:50:40,178 Epoch  24 Step:    36300 Batch Loss:     1.871893 Tokens per Sec:     7658, Lr: 0.000300\n",
      "2020-02-20 15:51:09,207 Epoch  24 Step:    36400 Batch Loss:     2.020421 Tokens per Sec:     7760, Lr: 0.000300\n",
      "2020-02-20 15:51:33,421 Epoch  24: total training loss 2640.82\n",
      "2020-02-20 15:51:33,421 EPOCH 25\n",
      "2020-02-20 15:51:38,282 Epoch  25 Step:    36500 Batch Loss:     1.380916 Tokens per Sec:     7233, Lr: 0.000300\n",
      "2020-02-20 15:52:07,143 Epoch  25 Step:    36600 Batch Loss:     1.699091 Tokens per Sec:     7551, Lr: 0.000300\n",
      "2020-02-20 15:52:35,780 Epoch  25 Step:    36700 Batch Loss:     1.813512 Tokens per Sec:     7590, Lr: 0.000300\n",
      "2020-02-20 15:53:04,754 Epoch  25 Step:    36800 Batch Loss:     1.879889 Tokens per Sec:     7619, Lr: 0.000300\n",
      "2020-02-20 15:53:33,617 Epoch  25 Step:    36900 Batch Loss:     1.529651 Tokens per Sec:     7609, Lr: 0.000300\n",
      "2020-02-20 15:54:02,465 Epoch  25 Step:    37000 Batch Loss:     1.654283 Tokens per Sec:     7575, Lr: 0.000300\n",
      "2020-02-20 15:55:30,642 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 15:55:30,642 Saving new checkpoint.\n",
      "2020-02-20 15:55:30,861 Example #0\n",
      "2020-02-20 15:55:30,862 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 15:55:30,862 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 15:55:30,862 \tHypothesis: Luwo chike Jehova e yo mowinjore , onego onyis ni chike Nyasaye ok owinjore .\n",
      "2020-02-20 15:55:30,862 Example #1\n",
      "2020-02-20 15:55:30,862 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 15:55:30,862 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 15:55:30,862 \tHypothesis: Bende , omiya luor matut ne wuonwa koda wiye . ”\n",
      "2020-02-20 15:55:30,862 Example #2\n",
      "2020-02-20 15:55:30,862 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 15:55:30,863 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 15:55:30,863 \tHypothesis: Nono gik ma Ndiko wacho nyalo konyo ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 15:55:30,863 Example #3\n",
      "2020-02-20 15:55:30,863 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 15:55:30,863 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 15:55:30,863 \tHypothesis: Noyudo osendiko e wi “ gik moko duto ma Nyasaye nowuoyo kuom dho jonabi machon . ”\n",
      "2020-02-20 15:55:30,863 Validation result at epoch  25, step    37000: bleu:  21.67, loss: 42305.6289, ppl:   5.5644, duration: 88.3971s\n",
      "2020-02-20 15:55:59,189 Epoch  25 Step:    37100 Batch Loss:     1.815181 Tokens per Sec:     7414, Lr: 0.000300\n",
      "2020-02-20 15:56:28,284 Epoch  25 Step:    37200 Batch Loss:     1.838013 Tokens per Sec:     7600, Lr: 0.000300\n",
      "2020-02-20 15:56:56,996 Epoch  25 Step:    37300 Batch Loss:     1.826828 Tokens per Sec:     7536, Lr: 0.000300\n",
      "2020-02-20 15:57:25,629 Epoch  25 Step:    37400 Batch Loss:     1.523698 Tokens per Sec:     7541, Lr: 0.000300\n",
      "2020-02-20 15:57:54,631 Epoch  25 Step:    37500 Batch Loss:     1.658947 Tokens per Sec:     7609, Lr: 0.000300\n",
      "2020-02-20 15:58:23,496 Epoch  25 Step:    37600 Batch Loss:     1.775190 Tokens per Sec:     7494, Lr: 0.000300\n",
      "2020-02-20 15:58:52,272 Epoch  25 Step:    37700 Batch Loss:     1.083689 Tokens per Sec:     7649, Lr: 0.000300\n",
      "2020-02-20 15:59:21,261 Epoch  25 Step:    37800 Batch Loss:     1.861390 Tokens per Sec:     7653, Lr: 0.000300\n",
      "2020-02-20 15:59:50,236 Epoch  25 Step:    37900 Batch Loss:     1.760268 Tokens per Sec:     7633, Lr: 0.000300\n",
      "2020-02-20 16:00:19,228 Epoch  25 Step:    38000 Batch Loss:     2.033724 Tokens per Sec:     7649, Lr: 0.000300\n",
      "2020-02-20 16:01:47,336 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:01:47,336 Saving new checkpoint.\n",
      "2020-02-20 16:01:47,557 Example #0\n",
      "2020-02-20 16:01:47,557 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:01:47,557 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:01:47,557 \tHypothesis: Luwo chike Jehova dwaro ni mondo onyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 16:01:47,557 Example #1\n",
      "2020-02-20 16:01:47,557 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:01:47,557 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:01:47,558 \tHypothesis: Bende , en gima miyo amiye luor kaka wuonwa koda wiye . ”\n",
      "2020-02-20 16:01:47,558 Example #2\n",
      "2020-02-20 16:01:47,558 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:01:47,558 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:01:47,558 \tHypothesis: Nono kaka Muma wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:01:47,558 Example #3\n",
      "2020-02-20 16:01:47,558 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:01:47,558 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:01:47,558 \tHypothesis: Noyudo osendiko weche duto ma Nyasaye ne wuoyo kuome e dho jonabi maler kuom kinde machon . ”\n",
      "2020-02-20 16:01:47,558 Validation result at epoch  25, step    38000: bleu:  21.64, loss: 42005.7266, ppl:   5.4971, duration: 88.3296s\n",
      "2020-02-20 16:01:48,159 Epoch  25: total training loss 2618.74\n",
      "2020-02-20 16:01:48,159 EPOCH 26\n",
      "2020-02-20 16:02:16,329 Epoch  26 Step:    38100 Batch Loss:     1.761483 Tokens per Sec:     7532, Lr: 0.000300\n",
      "2020-02-20 16:02:45,058 Epoch  26 Step:    38200 Batch Loss:     1.693808 Tokens per Sec:     7657, Lr: 0.000300\n",
      "2020-02-20 16:03:13,803 Epoch  26 Step:    38300 Batch Loss:     1.515896 Tokens per Sec:     7502, Lr: 0.000300\n",
      "2020-02-20 16:03:42,617 Epoch  26 Step:    38400 Batch Loss:     1.760625 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 16:04:11,016 Epoch  26 Step:    38500 Batch Loss:     0.938905 Tokens per Sec:     7395, Lr: 0.000300\n",
      "2020-02-20 16:04:39,951 Epoch  26 Step:    38600 Batch Loss:     1.577976 Tokens per Sec:     7537, Lr: 0.000300\n",
      "2020-02-20 16:05:08,704 Epoch  26 Step:    38700 Batch Loss:     1.703813 Tokens per Sec:     7629, Lr: 0.000300\n",
      "2020-02-20 16:05:37,617 Epoch  26 Step:    38800 Batch Loss:     1.682674 Tokens per Sec:     7560, Lr: 0.000300\n",
      "2020-02-20 16:06:06,358 Epoch  26 Step:    38900 Batch Loss:     1.234267 Tokens per Sec:     7645, Lr: 0.000300\n",
      "2020-02-20 16:06:35,344 Epoch  26 Step:    39000 Batch Loss:     1.931374 Tokens per Sec:     7689, Lr: 0.000300\n",
      "2020-02-20 16:08:03,511 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:08:03,512 Saving new checkpoint.\n",
      "2020-02-20 16:08:03,727 Example #0\n",
      "2020-02-20 16:08:03,728 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:08:03,728 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:08:03,728 \tHypothesis: Ka giikore winjo dwond Jehova , onego onyis ni chike Nyasaye ok owinjore .\n",
      "2020-02-20 16:08:03,728 Example #1\n",
      "2020-02-20 16:08:03,728 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:08:03,728 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:08:03,728 \tHypothesis: Bende , omiya luor matut ne en kaka wuonwa kod wiye . ”\n",
      "2020-02-20 16:08:03,728 Example #2\n",
      "2020-02-20 16:08:03,729 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:08:03,729 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:08:03,729 \tHypothesis: Nono ler mar gima Ndiko wacho nyalo konyo ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:08:03,729 Example #3\n",
      "2020-02-20 16:08:03,729 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:08:03,729 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:08:03,729 \tHypothesis: Noyudo osendiko e wi “ weche duto ma Nyasaye nowuoyo kuom dho jonabi maler mar ndalo machon . ”\n",
      "2020-02-20 16:08:03,729 Validation result at epoch  26, step    39000: bleu:  21.74, loss: 41955.6680, ppl:   5.4860, duration: 88.3847s\n",
      "2020-02-20 16:08:32,482 Epoch  26 Step:    39100 Batch Loss:     1.833119 Tokens per Sec:     7616, Lr: 0.000300\n",
      "2020-02-20 16:09:01,268 Epoch  26 Step:    39200 Batch Loss:     1.840668 Tokens per Sec:     7588, Lr: 0.000300\n",
      "2020-02-20 16:09:29,756 Epoch  26 Step:    39300 Batch Loss:     1.956726 Tokens per Sec:     7580, Lr: 0.000300\n",
      "2020-02-20 16:09:58,532 Epoch  26 Step:    39400 Batch Loss:     1.796959 Tokens per Sec:     7627, Lr: 0.000300\n",
      "2020-02-20 16:10:27,478 Epoch  26 Step:    39500 Batch Loss:     1.474928 Tokens per Sec:     7744, Lr: 0.000300\n",
      "2020-02-20 16:10:33,873 Epoch  26: total training loss 2605.51\n",
      "2020-02-20 16:10:33,873 EPOCH 27\n",
      "2020-02-20 16:10:56,508 Epoch  27 Step:    39600 Batch Loss:     1.582130 Tokens per Sec:     7580, Lr: 0.000300\n",
      "2020-02-20 16:11:25,268 Epoch  27 Step:    39700 Batch Loss:     1.310958 Tokens per Sec:     7513, Lr: 0.000300\n",
      "2020-02-20 16:11:54,123 Epoch  27 Step:    39800 Batch Loss:     1.736443 Tokens per Sec:     7625, Lr: 0.000300\n",
      "2020-02-20 16:12:23,131 Epoch  27 Step:    39900 Batch Loss:     1.639174 Tokens per Sec:     7540, Lr: 0.000300\n",
      "2020-02-20 16:12:52,007 Epoch  27 Step:    40000 Batch Loss:     1.594159 Tokens per Sec:     7664, Lr: 0.000300\n",
      "2020-02-20 16:14:20,170 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:14:20,170 Saving new checkpoint.\n",
      "2020-02-20 16:14:20,391 Example #0\n",
      "2020-02-20 16:14:20,391 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:14:20,391 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:14:20,391 \tHypothesis: Luwo chike Jehova e yo makare , onego onyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 16:14:20,391 Example #1\n",
      "2020-02-20 16:14:20,392 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:14:20,392 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:14:20,392 \tHypothesis: Bende , omiya luor matut kuom miye luor kaka wuonwa koda wiwa . ”\n",
      "2020-02-20 16:14:20,392 Example #2\n",
      "2020-02-20 16:14:20,392 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:14:20,392 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:14:20,392 \tHypothesis: Neno kaka Muma wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:14:20,392 Example #3\n",
      "2020-02-20 16:14:20,393 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:14:20,393 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:14:20,393 \tHypothesis: Noyudo osendiko e wi “ weche duto ma Nyasaye nowuoyo kuome e dho jonabi maler kuom kinde machon . ”\n",
      "2020-02-20 16:14:20,393 Validation result at epoch  27, step    40000: bleu:  21.32, loss: 41715.1719, ppl:   5.4327, duration: 88.3852s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 16:14:49,104 Epoch  27 Step:    40100 Batch Loss:     1.829319 Tokens per Sec:     7649, Lr: 0.000300\n",
      "2020-02-20 16:15:17,967 Epoch  27 Step:    40200 Batch Loss:     1.796776 Tokens per Sec:     7594, Lr: 0.000300\n",
      "2020-02-20 16:15:47,029 Epoch  27 Step:    40300 Batch Loss:     1.575827 Tokens per Sec:     7586, Lr: 0.000300\n",
      "2020-02-20 16:16:15,906 Epoch  27 Step:    40400 Batch Loss:     1.830190 Tokens per Sec:     7541, Lr: 0.000300\n",
      "2020-02-20 16:16:44,727 Epoch  27 Step:    40500 Batch Loss:     1.579970 Tokens per Sec:     7617, Lr: 0.000300\n",
      "2020-02-20 16:17:13,546 Epoch  27 Step:    40600 Batch Loss:     1.676199 Tokens per Sec:     7521, Lr: 0.000300\n",
      "2020-02-20 16:17:42,689 Epoch  27 Step:    40700 Batch Loss:     1.726659 Tokens per Sec:     7699, Lr: 0.000300\n",
      "2020-02-20 16:18:11,558 Epoch  27 Step:    40800 Batch Loss:     1.675419 Tokens per Sec:     7484, Lr: 0.000300\n",
      "2020-02-20 16:18:40,394 Epoch  27 Step:    40900 Batch Loss:     1.705130 Tokens per Sec:     7408, Lr: 0.000300\n",
      "2020-02-20 16:19:09,545 Epoch  27 Step:    41000 Batch Loss:     1.187504 Tokens per Sec:     7602, Lr: 0.000300\n",
      "2020-02-20 16:20:37,641 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:20:37,641 Saving new checkpoint.\n",
      "2020-02-20 16:20:37,860 Example #0\n",
      "2020-02-20 16:20:37,860 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:20:37,860 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:20:37,860 \tHypothesis: Luwo chik ma Jehova dwaro ni mondo ginyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 16:20:37,860 Example #1\n",
      "2020-02-20 16:20:37,861 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:20:37,861 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:20:37,861 \tHypothesis: Omiya luor bende miyo amiye luor kaka wuonwa gi Nyasaye . ”\n",
      "2020-02-20 16:20:37,861 Example #2\n",
      "2020-02-20 16:20:37,861 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:20:37,861 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:20:37,861 \tHypothesis: Neno kaka Muma wacho nyalo konyo ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:20:37,861 Example #3\n",
      "2020-02-20 16:20:37,862 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:20:37,862 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:20:37,862 \tHypothesis: Noyudo osendiko e wi “ liet mar gik moko duto ma Nyasaye nowuoyo kuome kokalo kuom jonabi maler ma ne tiyo . ”\n",
      "2020-02-20 16:20:37,862 Validation result at epoch  27, step    41000: bleu:  22.04, loss: 41422.9102, ppl:   5.3687, duration: 88.3159s\n",
      "2020-02-20 16:20:48,781 Epoch  27: total training loss 2575.73\n",
      "2020-02-20 16:20:48,782 EPOCH 28\n",
      "2020-02-20 16:21:06,622 Epoch  28 Step:    41100 Batch Loss:     1.798172 Tokens per Sec:     7491, Lr: 0.000300\n",
      "2020-02-20 16:21:35,147 Epoch  28 Step:    41200 Batch Loss:     1.775519 Tokens per Sec:     7441, Lr: 0.000300\n",
      "2020-02-20 16:22:03,839 Epoch  28 Step:    41300 Batch Loss:     1.444540 Tokens per Sec:     7504, Lr: 0.000300\n",
      "2020-02-20 16:22:32,533 Epoch  28 Step:    41400 Batch Loss:     1.862024 Tokens per Sec:     7554, Lr: 0.000300\n",
      "2020-02-20 16:23:01,616 Epoch  28 Step:    41500 Batch Loss:     1.405001 Tokens per Sec:     7729, Lr: 0.000300\n",
      "2020-02-20 16:23:30,363 Epoch  28 Step:    41600 Batch Loss:     1.625368 Tokens per Sec:     7482, Lr: 0.000300\n",
      "2020-02-20 16:23:59,021 Epoch  28 Step:    41700 Batch Loss:     1.713350 Tokens per Sec:     7432, Lr: 0.000300\n",
      "2020-02-20 16:24:27,949 Epoch  28 Step:    41800 Batch Loss:     1.636419 Tokens per Sec:     7589, Lr: 0.000300\n",
      "2020-02-20 16:24:56,914 Epoch  28 Step:    41900 Batch Loss:     1.398697 Tokens per Sec:     7678, Lr: 0.000300\n",
      "2020-02-20 16:25:25,811 Epoch  28 Step:    42000 Batch Loss:     1.813509 Tokens per Sec:     7645, Lr: 0.000300\n",
      "2020-02-20 16:26:53,604 Example #0\n",
      "2020-02-20 16:26:53,604 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:26:53,604 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:26:53,605 \tHypothesis: Luwo chike Jehova dwaro ni mondo onyis ni chike Nyasaye ok owinjore .\n",
      "2020-02-20 16:26:53,605 Example #1\n",
      "2020-02-20 16:26:53,605 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:26:53,605 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:26:53,605 \tHypothesis: Bende , omiya thuolo mar miyo luor kaka wuonwa kod wiye otegno . ”\n",
      "2020-02-20 16:26:53,605 Example #2\n",
      "2020-02-20 16:26:53,605 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:26:53,605 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:26:53,605 \tHypothesis: Neno kaka Muma wacho nyalo konyo ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:26:53,605 Example #3\n",
      "2020-02-20 16:26:53,606 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:26:53,606 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:26:53,606 \tHypothesis: Noyudo osendiko e wi “ weche duto ma Nyasaye nowuoyo kuome kokalo kuom jonabi maler mag higini machon . ”\n",
      "2020-02-20 16:26:53,606 Validation result at epoch  28, step    42000: bleu:  22.21, loss: 41490.4336, ppl:   5.3834, duration: 87.7938s\n",
      "2020-02-20 16:27:22,751 Epoch  28 Step:    42100 Batch Loss:     1.777676 Tokens per Sec:     7661, Lr: 0.000300\n",
      "2020-02-20 16:27:51,485 Epoch  28 Step:    42200 Batch Loss:     1.654832 Tokens per Sec:     7571, Lr: 0.000300\n",
      "2020-02-20 16:28:20,580 Epoch  28 Step:    42300 Batch Loss:     1.638294 Tokens per Sec:     7781, Lr: 0.000300\n",
      "2020-02-20 16:28:49,416 Epoch  28 Step:    42400 Batch Loss:     1.761941 Tokens per Sec:     7631, Lr: 0.000300\n",
      "2020-02-20 16:29:18,212 Epoch  28 Step:    42500 Batch Loss:     1.954315 Tokens per Sec:     7629, Lr: 0.000300\n",
      "2020-02-20 16:29:34,278 Epoch  28: total training loss 2562.86\n",
      "2020-02-20 16:29:34,279 EPOCH 29\n",
      "2020-02-20 16:29:46,915 Epoch  29 Step:    42600 Batch Loss:     1.826073 Tokens per Sec:     7389, Lr: 0.000300\n",
      "2020-02-20 16:30:15,572 Epoch  29 Step:    42700 Batch Loss:     1.895255 Tokens per Sec:     7517, Lr: 0.000300\n",
      "2020-02-20 16:30:44,511 Epoch  29 Step:    42800 Batch Loss:     1.770706 Tokens per Sec:     7666, Lr: 0.000300\n",
      "2020-02-20 16:31:13,390 Epoch  29 Step:    42900 Batch Loss:     1.765434 Tokens per Sec:     7636, Lr: 0.000300\n",
      "2020-02-20 16:31:42,077 Epoch  29 Step:    43000 Batch Loss:     1.890643 Tokens per Sec:     7591, Lr: 0.000300\n",
      "2020-02-20 16:33:10,168 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:33:10,168 Saving new checkpoint.\n",
      "2020-02-20 16:33:10,395 Example #0\n",
      "2020-02-20 16:33:10,396 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:33:10,396 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:33:10,396 \tHypothesis: Luwo chike Jehova dwaro ni ginyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 16:33:10,396 Example #1\n",
      "2020-02-20 16:33:10,396 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:33:10,396 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:33:10,396 \tHypothesis: Bende , omiya luor matut kaka wuonwa kod wiye . ”\n",
      "2020-02-20 16:33:10,396 Example #2\n",
      "2020-02-20 16:33:10,397 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:33:10,397 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:33:10,397 \tHypothesis: Nono gik ma Ndiko wacho nyalo konyowa ng’eyo kaka chunywa dwaro .\n",
      "2020-02-20 16:33:10,397 Example #3\n",
      "2020-02-20 16:33:10,397 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:33:10,397 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:33:10,397 \tHypothesis: Noyudo osendiko e wi “ weche duto ma Nyasaye nowacho kokalo kuom jonabi maler mag higini machon . ”\n",
      "2020-02-20 16:33:10,397 Validation result at epoch  29, step    43000: bleu:  22.42, loss: 41402.2656, ppl:   5.3642, duration: 88.3194s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-02-20 16:33:39,037 Epoch  29 Step:    43100 Batch Loss:     1.562337 Tokens per Sec:     7495, Lr: 0.000300\n",
      "2020-02-20 16:34:07,539 Epoch  29 Step:    43200 Batch Loss:     1.512897 Tokens per Sec:     7540, Lr: 0.000300\n",
      "2020-02-20 16:34:36,579 Epoch  29 Step:    43300 Batch Loss:     1.912865 Tokens per Sec:     7742, Lr: 0.000300\n",
      "2020-02-20 16:35:05,179 Epoch  29 Step:    43400 Batch Loss:     1.795569 Tokens per Sec:     7543, Lr: 0.000300\n",
      "2020-02-20 16:35:34,011 Epoch  29 Step:    43500 Batch Loss:     1.239226 Tokens per Sec:     7626, Lr: 0.000300\n",
      "2020-02-20 16:36:02,734 Epoch  29 Step:    43600 Batch Loss:     1.947909 Tokens per Sec:     7432, Lr: 0.000300\n",
      "2020-02-20 16:36:31,516 Epoch  29 Step:    43700 Batch Loss:     1.863791 Tokens per Sec:     7587, Lr: 0.000300\n",
      "2020-02-20 16:37:00,585 Epoch  29 Step:    43800 Batch Loss:     1.787845 Tokens per Sec:     7634, Lr: 0.000300\n",
      "2020-02-20 16:37:29,387 Epoch  29 Step:    43900 Batch Loss:     1.920231 Tokens per Sec:     7677, Lr: 0.000300\n",
      "2020-02-20 16:37:57,968 Epoch  29 Step:    44000 Batch Loss:     1.613283 Tokens per Sec:     7532, Lr: 0.000300\n",
      "2020-02-20 16:39:25,983 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:39:25,983 Saving new checkpoint.\n",
      "2020-02-20 16:39:26,198 Example #0\n",
      "2020-02-20 16:39:26,199 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:39:26,199 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:39:26,199 \tHypothesis: Luwo chike Jehova dwaro ni mondo ginyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 16:39:26,199 Example #1\n",
      "2020-02-20 16:39:26,199 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:39:26,199 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:39:26,199 \tHypothesis: Omiya thuolo mar miye luor kaka wuonwa kendo tayo winjruokna gi Nyasaye . ”\n",
      "2020-02-20 16:39:26,200 Example #2\n",
      "2020-02-20 16:39:26,200 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:39:26,200 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:39:26,200 \tHypothesis: Neno kaka Muma wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:39:26,200 Example #3\n",
      "2020-02-20 16:39:26,200 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:39:26,200 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:39:26,200 \tHypothesis: Nondiko e wi “ weche duto ma Nyasaye nowacho kokalo kuom jonabi maler kuom kinde machon . ”\n",
      "2020-02-20 16:39:26,200 Validation result at epoch  29, step    44000: bleu:  21.92, loss: 41178.2109, ppl:   5.3156, duration: 88.2323s\n",
      "2020-02-20 16:39:48,441 Epoch  29: total training loss 2546.09\n",
      "2020-02-20 16:39:48,441 EPOCH 30\n",
      "2020-02-20 16:39:55,621 Epoch  30 Step:    44100 Batch Loss:     1.842117 Tokens per Sec:     7399, Lr: 0.000300\n",
      "2020-02-20 16:40:24,504 Epoch  30 Step:    44200 Batch Loss:     1.342121 Tokens per Sec:     7623, Lr: 0.000300\n",
      "2020-02-20 16:40:53,230 Epoch  30 Step:    44300 Batch Loss:     1.576577 Tokens per Sec:     7583, Lr: 0.000300\n",
      "2020-02-20 16:41:22,240 Epoch  30 Step:    44400 Batch Loss:     1.735350 Tokens per Sec:     7642, Lr: 0.000300\n",
      "2020-02-20 16:41:51,012 Epoch  30 Step:    44500 Batch Loss:     1.906591 Tokens per Sec:     7496, Lr: 0.000300\n",
      "2020-02-20 16:42:19,766 Epoch  30 Step:    44600 Batch Loss:     1.563841 Tokens per Sec:     7569, Lr: 0.000300\n",
      "2020-02-20 16:42:48,494 Epoch  30 Step:    44700 Batch Loss:     1.597712 Tokens per Sec:     7567, Lr: 0.000300\n",
      "2020-02-20 16:43:17,344 Epoch  30 Step:    44800 Batch Loss:     1.467246 Tokens per Sec:     7521, Lr: 0.000300\n",
      "2020-02-20 16:43:45,937 Epoch  30 Step:    44900 Batch Loss:     1.723284 Tokens per Sec:     7495, Lr: 0.000300\n",
      "2020-02-20 16:44:14,470 Epoch  30 Step:    45000 Batch Loss:     1.647184 Tokens per Sec:     7650, Lr: 0.000300\n",
      "2020-02-20 16:45:42,572 Hooray! New best validation result [ppl]!\n",
      "2020-02-20 16:45:42,572 Saving new checkpoint.\n",
      "2020-02-20 16:45:42,798 Example #0\n",
      "2020-02-20 16:45:42,798 \tSource:     Their own willing obedience to what Jehovah requires should demonstrate that God’s rules are reasonable .\n",
      "2020-02-20 16:45:42,798 \tReference:  Ka gin giwegi giikore luwo gima Jehova dwaro , mano biro nyiso ni chike Nyasaye gin chike mowinjore kendo inyalo makgi .\n",
      "2020-02-20 16:45:42,798 \tHypothesis: Luwo chike Jehova e yo ma Jehova dwaro ni mondo onyis ni chike Nyasaye owinjore .\n",
      "2020-02-20 16:45:42,798 Example #1\n",
      "2020-02-20 16:45:42,798 \tSource:     It also gives me more reason to respect him as my father and spiritual head . ”\n",
      "2020-02-20 16:45:42,799 \tReference:  Mano bende miyo amedo miye luor kaka wuonwa kendo kaka ng’at matayowa e weche mag lamo . ”\n",
      "2020-02-20 16:45:42,799 \tHypothesis: Bende , omiyo amiyo luor kaka wuonwa kod wiye . ”\n",
      "2020-02-20 16:45:42,799 Example #2\n",
      "2020-02-20 16:45:42,799 \tSource:     Seeing ourselves in the light of what the Scriptures say can indeed be a powerful aid in identifying the intentions of the heart .\n",
      "2020-02-20 16:45:42,799 \tReference:  Ka wanonore wawegi kaluwore gi gima Ndiko wacho , mano nyalo konyowa ng’eyo paro manie chunywa .\n",
      "2020-02-20 16:45:42,799 \tHypothesis: Neno kaka Muma wacho nyalo bedo gima duong ’ ahinya e fwenyo chuny ng’ato .\n",
      "2020-02-20 16:45:42,799 Example #3\n",
      "2020-02-20 16:45:42,799 \tSource:     He had written about the “ restoration of all things of which God spoke through the mouth of his holy prophets of old time . ”\n",
      "2020-02-20 16:45:42,799 \tReference:  Noyudo Dunn osendiko wach “ loso gik moko duto odok makare , kaka Nyasaye nowacho gi dho jonabi duto maler nyaka a chakruok piny . ”\n",
      "2020-02-20 16:45:42,799 \tHypothesis: Noyudo osendiko e wi “ weche duto ma Nyasaye ne wuoyo kuome kokalo kuom jonabi maler ma ne okoro chon . ”\n",
      "2020-02-20 16:45:42,800 Validation result at epoch  30, step    45000: bleu:  22.35, loss: 41098.9102, ppl:   5.2986, duration: 88.3286s\n",
      "2020-02-20 16:46:11,424 Epoch  30 Step:    45100 Batch Loss:     1.743886 Tokens per Sec:     7400, Lr: 0.000300\n",
      "2020-02-20 16:46:40,731 Epoch  30 Step:    45200 Batch Loss:     1.809998 Tokens per Sec:     7796, Lr: 0.000300\n",
      "2020-02-20 16:47:09,258 Epoch  30 Step:    45300 Batch Loss:     1.818204 Tokens per Sec:     7525, Lr: 0.000300\n",
      "2020-02-20 16:47:38,422 Epoch  30 Step:    45400 Batch Loss:     1.736148 Tokens per Sec:     7673, Lr: 0.000300\n",
      "2020-02-20 16:48:07,190 Epoch  30 Step:    45500 Batch Loss:     1.416873 Tokens per Sec:     7698, Lr: 0.000300\n",
      "2020-02-20 16:48:34,725 Epoch  30: total training loss 2530.09\n",
      "2020-02-20 16:48:34,725 Training ended after  30 epochs.\n",
      "2020-02-20 16:48:34,725 Best validation result at step    45000:   5.30 ppl.\n",
      "2020-02-20 16:49:17,978  dev bleu:  23.22 [Beam search decoding with beam size = 5 and alpha = 1.0]\n",
      "2020-02-20 16:49:17,979 Translations saved to: models/enluo_transformer/00045000.hyps.dev\n",
      "2020-02-20 16:50:37,458 test bleu:  32.64 [Beam search decoding with beam size = 5 and alpha = 1.0]\n",
      "2020-02-20 16:50:37,460 Translations saved to: models/enluo_transformer/00045000.hyps.test\n"
     ]
    }
   ],
   "source": [
    "# Train the model\n",
    "# You can press Ctrl-C to stop. And then run the next cell to save your checkpoints! \n",
    "! cd ../../joeynmt; python3 -m joeynmt train configs/transformer_$src$tgt.yaml"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "MBoDS09JM807"
   },
   "outputs": [],
   "source": [
    "# Copy the created models from the notebook storage to google drive for persistant storage \n",
    "!cp -r joeynmt/models/${src}${tgt}_transformer/* \"$gdrive_path/models/${src}${tgt}_transformer/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "n94wlrCjVc17"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Steps: 1000\tLoss: 98175.97656\tPPL: 53.68433\tbleu: 1.00511\tLR: 0.00030000\t*\r\n",
      "Steps: 2000\tLoss: 83177.78906\tPPL: 29.21336\tbleu: 2.74320\tLR: 0.00030000\t*\r\n",
      "Steps: 3000\tLoss: 75435.21094\tPPL: 21.33823\tbleu: 4.85840\tLR: 0.00030000\t*\r\n",
      "Steps: 4000\tLoss: 69503.42969\tPPL: 16.77417\tbleu: 7.70259\tLR: 0.00030000\t*\r\n",
      "Steps: 5000\tLoss: 65853.02344\tPPL: 14.46510\tbleu: 9.55595\tLR: 0.00030000\t*\r\n",
      "Steps: 6000\tLoss: 63031.21875\tPPL: 12.90035\tbleu: 10.99594\tLR: 0.00030000\t*\r\n",
      "Steps: 7000\tLoss: 60268.74609\tPPL: 11.53260\tbleu: 11.95757\tLR: 0.00030000\t*\r\n",
      "Steps: 8000\tLoss: 58475.81250\tPPL: 10.72349\tbleu: 12.97758\tLR: 0.00030000\t*\r\n",
      "Steps: 9000\tLoss: 57032.02734\tPPL: 10.11339\tbleu: 13.53204\tLR: 0.00030000\t*\r\n",
      "Steps: 10000\tLoss: 54667.55078\tPPL: 9.18829\tbleu: 15.21627\tLR: 0.00030000\t*\r\n",
      "Steps: 11000\tLoss: 53643.98438\tPPL: 8.81454\tbleu: 15.05572\tLR: 0.00030000\t*\r\n",
      "Steps: 12000\tLoss: 52487.89453\tPPL: 8.41065\tbleu: 15.88412\tLR: 0.00030000\t*\r\n",
      "Steps: 13000\tLoss: 51491.73047\tPPL: 8.07751\tbleu: 16.26519\tLR: 0.00030000\t*\r\n",
      "Steps: 14000\tLoss: 50492.83203\tPPL: 7.75670\tbleu: 17.06413\tLR: 0.00030000\t*\r\n",
      "Steps: 15000\tLoss: 49596.87500\tPPL: 7.47981\tbleu: 18.29802\tLR: 0.00030000\t*\r\n",
      "Steps: 16000\tLoss: 49577.44141\tPPL: 7.47391\tbleu: 17.63687\tLR: 0.00030000\t*\r\n",
      "Steps: 17000\tLoss: 48477.73438\tPPL: 7.14778\tbleu: 17.78224\tLR: 0.00030000\t*\r\n",
      "Steps: 18000\tLoss: 47651.57812\tPPL: 6.91217\tbleu: 18.38615\tLR: 0.00030000\t*\r\n",
      "Steps: 19000\tLoss: 47321.08984\tPPL: 6.82011\tbleu: 18.75187\tLR: 0.00030000\t*\r\n",
      "Steps: 20000\tLoss: 46756.95312\tPPL: 6.66578\tbleu: 19.03500\tLR: 0.00030000\t*\r\n",
      "Steps: 21000\tLoss: 46212.31641\tPPL: 6.52011\tbleu: 19.73430\tLR: 0.00030000\t*\r\n",
      "Steps: 22000\tLoss: 46018.47656\tPPL: 6.46903\tbleu: 19.07405\tLR: 0.00030000\t*\r\n",
      "Steps: 23000\tLoss: 45625.03516\tPPL: 6.36659\tbleu: 19.67849\tLR: 0.00030000\t*\r\n",
      "Steps: 24000\tLoss: 45200.48047\tPPL: 6.25787\tbleu: 20.11382\tLR: 0.00030000\t*\r\n",
      "Steps: 25000\tLoss: 44906.89844\tPPL: 6.18377\tbleu: 19.90823\tLR: 0.00030000\t*\r\n",
      "Steps: 26000\tLoss: 44817.06641\tPPL: 6.16128\tbleu: 19.35355\tLR: 0.00030000\t*\r\n",
      "Steps: 27000\tLoss: 44218.17578\tPPL: 6.01338\tbleu: 20.43616\tLR: 0.00030000\t*\r\n",
      "Steps: 28000\tLoss: 44157.72656\tPPL: 5.99865\tbleu: 20.51112\tLR: 0.00030000\t*\r\n",
      "Steps: 29000\tLoss: 43801.03516\tPPL: 5.91246\tbleu: 20.96693\tLR: 0.00030000\t*\r\n",
      "Steps: 30000\tLoss: 43687.10547\tPPL: 5.88520\tbleu: 21.00739\tLR: 0.00030000\t*\r\n",
      "Steps: 31000\tLoss: 43309.37500\tPPL: 5.79569\tbleu: 20.82572\tLR: 0.00030000\t*\r\n",
      "Steps: 32000\tLoss: 43100.05078\tPPL: 5.74668\tbleu: 21.35273\tLR: 0.00030000\t*\r\n",
      "Steps: 33000\tLoss: 42939.68750\tPPL: 5.70941\tbleu: 20.98994\tLR: 0.00030000\t*\r\n",
      "Steps: 34000\tLoss: 42751.12891\tPPL: 5.66590\tbleu: 21.51582\tLR: 0.00030000\t*\r\n",
      "Steps: 35000\tLoss: 42456.41797\tPPL: 5.59856\tbleu: 21.60048\tLR: 0.00030000\t*\r\n",
      "Steps: 36000\tLoss: 42349.58203\tPPL: 5.57435\tbleu: 21.19294\tLR: 0.00030000\t*\r\n",
      "Steps: 37000\tLoss: 42305.62891\tPPL: 5.56441\tbleu: 21.66766\tLR: 0.00030000\t*\r\n",
      "Steps: 38000\tLoss: 42005.72656\tPPL: 5.49712\tbleu: 21.63990\tLR: 0.00030000\t*\r\n",
      "Steps: 39000\tLoss: 41955.66797\tPPL: 5.48597\tbleu: 21.74473\tLR: 0.00030000\t*\r\n",
      "Steps: 40000\tLoss: 41715.17188\tPPL: 5.43270\tbleu: 21.32302\tLR: 0.00030000\t*\r\n",
      "Steps: 41000\tLoss: 41422.91016\tPPL: 5.36866\tbleu: 22.04028\tLR: 0.00030000\t*\r\n",
      "Steps: 42000\tLoss: 41490.43359\tPPL: 5.38339\tbleu: 22.21160\tLR: 0.00030000\t\r\n",
      "Steps: 43000\tLoss: 41402.26562\tPPL: 5.36417\tbleu: 22.42439\tLR: 0.00030000\t*\r\n",
      "Steps: 44000\tLoss: 41178.21094\tPPL: 5.31563\tbleu: 21.92493\tLR: 0.00030000\t*\r\n",
      "Steps: 45000\tLoss: 41098.91016\tPPL: 5.29855\tbleu: 22.34637\tLR: 0.00030000\t*\r\n"
     ]
    }
   ],
   "source": [
    "# Output our validation accuracy\n",
    "! cat \"../../joeynmt/models/${src}${tgt}_transformer/validations.txt\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "66WhRE9lIhoD"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:541: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:542: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:543: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:544: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:545: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "/home/espoir_mur_gmail_com/.local/lib/python3.6/site-packages/tensorboard/compat/tensorflow_stub/dtypes.py:550: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n",
      "2020-02-20 16:51:25,414 -  dev bleu:  23.22 [Beam search decoding with beam size = 5 and alpha = 1.0]\n",
      "2020-02-20 16:52:45,303 - test bleu:  32.64 [Beam search decoding with beam size = 5 and alpha = 1.0]\n"
     ]
    }
   ],
   "source": [
    "# Test our model\n",
    "! cd ../../joeynmt; python3 -m joeynmt test \"models/${src}${tgt}_transformer/config.yaml\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "starter_notebook.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}