A short question about the moses flag "-output-factors":
I tried to train a factored model that translates some source factors to
one target language factor only (which is not the first one). I always
ended up with segmentation faults after translating the first sentence
and it took me some time to find out that I have to specify the
output-factor with the argument above. Why is this necessary if I only
output one factor (even though it's not factor 0)? Maybe a useful error
message would be helpful to detect this error. (I think Moses crashed at
Factor.cpp:47 without the additional argument)
Another thing: I had a problem using the phrase scoring script when
using my own phrase extraction (giving me segmentation faults again).
The problem is that my phrases are not really compatible with the GIZA++
word alignments I guess and there might be some mismatch between word
alignments in my phrases and the lexicon files. Would it be possible to
add the following 2 lines to score.cpp before line 355 just for an
additional sanity check (it doesn't hurt anything else I believe)
if (alignedToE.size() < i) return false;
if (other.alignedToE.size() < i) return false;
Thanks!
--
Jörg
*******/\/\/\/\/\/\/\/\/\/\/\******************************************
Jörg Tiedemann [email protected]
Visiting Professor http://stp.lingfil.uu.se/~joerg/
Dep. of Linguistics and Philology
Uppsala University tel: (018) 471 1412
Box 635, SE-751 26 Uppsala/SWEDEN fax: (018) 471 1094
*********************************/\/\/\/\/\/\/\/\/\/\/\****************
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
