hi shaohua
it may be that you've specified the using the wrong LM implementation. 3
implementations are supported in Moses: SRI, IRST, KenLM.
if you binarize with the tools from 1 of these, you must specifiy that
the decoder runs with that implementation.
therefore, exactly how did you create & binary the LM, and can you
please attach the moses.ini file
hieu
On 08/12/2011 18:41, ??? wrote:
Dear all,
I'm following the moses step by step tutorial to train a
phrase-based mt model, when I ran the "mert-moses.pl
<http://mert-moses.pl>" command, the process stop after a little time.
And the mert.log file ends like this:
Loading lexical distortion models...
112 have 1 models
113 Creating lexical reordering...
114 weights: 0.111 0.111 0.111 0.111 0.111 0.111
115 Loading table into memory...done.
116 Created lexical orientation reordering
117
/home/yangshaohua/experiment/no_modified_character/work1/lm/data.lm:
line 13: warning: non-zero probability for <unk> in closed-vocabulary LM
118 Segmentation fault
119 Exit code: 139
120 The decoder died. CONFIG WAS -w 0.000000 -lm 0.111111 -d 0.111111
0.111111 0.111111 0.111111 0.111111 0.111111 0.111111 -tm 0.033333
0.022222 0.033333 0.022222 0.000000
I don't think that this problem is caused by limited memory because
when it stops, the system still has a lot of memory. What else
problems may be for such a situation?
*Best Regards,*
*ShaoHua yang*
*
*
_______________________________________________
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
