another solution for the use of binary tables is to leave the original plain text tables in the directory as well. tuning will still use the binary ones if they are there.
about the memory problems: maybe switching off caching would help you? add this to the config file: [use-persistent-cache] 0 that usually does it for me. (but it will slow down things quite a lot!) jorg Attila Zséder wrote: > Hi Barry, > > Yes, it's working now, thank you! > > ( Although, i ran into another problem, same as here: > http://thread.gmane.org/gmane.comp.nlp.moses.user/1511/focus=1517 but > this problem has been already discussed on that thread) > > Br, > Attila > > On Thu, Mar 12, 2009 at 7:08 PM, Barry Haddow <[email protected]> wrote: >> Hi Attila >> >> Could you try changing line 1152 of mert-moses-new.pl from >> >> if (! -s $fn && ! -s "$fn.gz" && ! -s "$fn.binphr.idx") { >> >> to >> >> if (! -s $fn && ! -s "$fn.gz" && ! -s "$fn.binphr.idx" >> && ! -s "$fn.binlexr.idx") { >> >> and see if that fixes the problem? >> >> regards >> Barry >> >> On Thursday 12 March 2009 15:46, Attila Zséder wrote: >>> Hello, >>> >>> I have successfully built a translation model, and than i binarized >>> both the phrase table and the reordering table. I followed the >>> instructions found here: >>> http://www.statmt.org/moses/?n=Moses.AdvancedFeatures#ntoc2 . When i >>> try to decode, everything works fine, I get a translation. (So i >>> assume that i managed to change model/moses.ini in order to use binary >>> tables. ) >>> Btw, the actual lines are: >>> [ttable-file] >>> 0 0 5 /big3/zseder/mt/model/phrase-table >>> [...] >>> [distortion-file] >>> 0-0 msd-bidirectional-fe 6 /big3/zseder/mt/model/reordering-table >>> >>> So, decoding goes fine. But if i want to use mert-moses-new.pl to >>> tuning phase, i got the following problem: >>> moses.ini:31:File does not exist or empty: >>> /big3/zseder/mt/model/reordering-table >>> (my invoking command was: >>> /home/zseder/.local/bin/moses-scripts/scripts-20081004-1246/training/mert-m >>> oses-new.pl /big3/zseder/mt/temp/tuning_lowercased.hu >>> /big3/zseder/mt/temp/tuning_lowercased.en >>> /home/zseder/moses/moses-cmd/src/moses /big3/zseder/mt/model/moses.ini >>> --working-dir /big3/zseder/mt/tuning --rootdir >>> /home/zseder/.local/bin/moses-scripts/scripts-20081004-1246 >>> --mertdir=/home/zseder/moses/mert/ >>> ) >>> >>> Im sure that i use the same moses.ini for the successful decoding and >>> for unsuccessful tuning. >>> So, my question is: how could i tell to mert-moses-new that i want to >>> use a binary reordering table? Is it possible at all? I did not find >>> anything in the archives. >>> >>> Thanks in advance! >>> >>> Br, >>> Attila >>> _______________________________________________ >>> Moses-support mailing list >>> [email protected] >>> http://mailman.mit.edu/mailman/listinfo/moses-support >> -- >> The University of Edinburgh is a charitable body, registered in >> Scotland, with registration number SC005336. >> >> > > _______________________________________________ > Moses-support mailing list > [email protected] > http://mailman.mit.edu/mailman/listinfo/moses-support -- Jörg ***********/\/\/\/\/\/\/\/\/\/\/\************************************ ** Jörg Tiedemann [email protected] ** ** Alfa-Informatica http://www.let.rug.nl/~tiedeman ** ** Rijksuniversiteit Groningen Harmoniegebouw, room 1311-429 ** ** Postbus 716 phone: +31 (0)50-363 5935 ** ** 9700 AS Groningen fax: +31 (0)50-363 6855 ** *************************************/\/\/\/\/\/\/\/\/\/\/\********** _______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
