Hi Vi, You need to look into phrase table. I suspect that "sessions ||| ..." does not appear in the phrase table.
Nguyen Pham Thi Anh Vi wrote: > Hi all, > I've trained a model with English-Vietnameses corpus. After training, I > received the lexical file* lex.0-0.n2f*. I found in this file lines as > following : > > ... > sweet-bay câu 0.0012255 > *sessions *vào 0.0001057 > *sessions *họp 0.0042553 > *sessions *vì 0.0003054 > *sessions *phiên_toà 0.0312500 > *sessions *2 0.0031949 > *sessions *hàng 0.0004948 > *sessions *vậy_nên 0.1000000 > *sessions *buổi 0.0052174 > *sessions *ban_sáng 0.2500000 > *sessions *học 0.0008078 > *sessions *quý 0.0028169 > psalmodic hát 0.0024752 > ... > > But when I ran decoding with "*sessions*" word, I received the result: > > BEST TRANSLATION: sessions|UNK|UNK|UNK [1] [total=-110.537] <<0.000, > -1.000, -100.000, 0.000, 0.000, 0.000, 0.000, 0.000, 0.000, -23.073, 0.000, > 0.000, 0.000, 0.000, 0.000>> > sessions > reset caches > htable class statistics > size 800000 keys 2 acc 4 coll 0 used memory 3126Kb > htable class statistics > size 400000 keys 1 acc 2 coll 0 used memory 1563Kb > htable class statistics > size 400000 keys 3 acc 14 coll 0 used memory 1563Kb > htable class statistics > size 400000 keys 0 acc 0 coll 0 used memory 1563Kb > htable class statistics > size 400000 keys 0 acc 0 coll 0 used memory 1563Kb > reset mmap > Translation took 0.260 seconds > Finished translating > > I do not know why *sessions *was translated as UNK word. Could you please > help me? Can I fix this? And how? > Thanks for helping me! > > > Vi, > > ------------------------------------------------------------------------ > > _______________________________________________ > Moses-support mailing list > [email protected] > http://mailman.mit.edu/mailman/listinfo/moses-support > _______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
