Hello: yes, I have used relative paths in "-root-dir" and "-corpus" parameters
$SCRIPTS_ROOTDIR/training/train-model.perl \ -scripts-root-dir $SCRIPTS_ROOTDIR \ -root-dir $f-$e \ -corpus data/tagged/train.cln2 \ -f $f -e $e \ -alignment grow-diag-final-and \ -reordering msd-bidirectional-fe \ -lm 0:$Gram:$WD/lm/train.$e.qblm.mm:1 \ -lm 1:$Gram:$WD/lm/train-pos.$e.qblm.mm:1 \ --translation-factors 0-0,1 I'll do like you say. Thank you. 2011/10/24 Eleftherios Avramidis <[email protected]> > ** > ln may be failing because it creates link that points to a relative path... > the problem is that this relative path is seen from where your script is > being executed and not from where it is actually placed. > > Have you specified relative paths as parameters? I would just simply try to > specify absolute paths for all in/out-coming files > > Lefteris > > > > On 24/10/11 16:00, Alexandre Helle wrote: > > Hello, > > the problem with the zero-lenght is solved, it was a problem in the tagging > of the japanese, solved like you say. > > But continues dying with: > > (4) generate lexical translation table 0-0,1 @ Mon Oct 24 14:15:47 CEST >> 2011 >> (4) [0-0,1] generate lexical translation table @ Mon Oct 24 14:15:47 CEST >> 2011 >> (1.0.5) reducing factors to produce ja-en/model/aligned.0.ja @ Mon Oct 24 >> 14:15:47 CEST 2011 >> ..........(100000)..........(200000)..........(300000).... >> (1.0.5) reducing factors to produce ja-en/model/aligned.0,1.en @ Mon Oct >> 24 14:15:59 CEST 2011 >> Executing: ln -s data/tagged/train.cln2.en ja-en/model/aligned.0,1.en >> >> (ja-en/model/aligned.0.ja,ja-en/model/aligned.0,1.en,ja-en/model/lex.0-0,1) >> ERROR: Can't read ja-en/model/aligned.0,1.en at >> /Data/moses/tools/bin/moses-scripts/training/train-model.perl line 1132. >> FILE: ja-en/model/aligned.0,1.en >> > > Fails creating the link in "ln -s data/tagged/train.cln2.en > ja-en/model/aligned.0,1.en", but the file "data/tagged/train.cln2.en" exists > and has data. > > > Best regards and thank you. > > 2011/10/24 Eleftherios Avramidis <[email protected]> > >> Hi Alexandre, >> >> Giza doesn't like it when you give it sentence pairs with no words. You >> should filter out these sentences before >> >> Lefteris >> >> >> On 22/10/11 14:43, Alexandre Helle wrote: >> >> Hello, >> >> I'm training factored models but I get the next error: >> >> $SCRIPTS_ROOTDIR/training/train-model.perl -scripts-root-dir >> $SCRIPTS_ROOTDIR -root-dir $f-$e -corpus data/tagged/train.cln -f $f -e $e >> -lm 0:$Gram:$WD/lm/train.$e.qblm.mm:1 -lm 1:$Gram:$WD/lm/train-pos.$ >> e.qblm.mm:1 --translation-factors 0-0,1 >> (...) >> Reading more sentence pairs into memory ... >> ERROR: Forbidden zero sentence length 0 >> ERROR: Forbidden zero sentence length 0 >> ERROR: Forbidden zero sentence length 0 >> ERROR: Forbidden zero sentence length 0 >> ERROR: Forbidden zero sentence length 0 >> (...) >> ERROR: Forbidden zero sentence length 0 >> ERROR: Forbidden zero sentence length 0 >> ERROR: Execution of: /Data/moses/tools/bin/GIZA++ -CoocurrenceFile >> en-ja/giza.ja-en/ja-en.cooc -c en-ja/corpus/ja-en-int-train.snt -m1 5 -m2 0 >> -m3 3 -m4 3 -model1dumpfrequency 1 -model4smoothfactor 0.4 -nodumps 1 >> -nsmooth 4 -o en-ja/giza.ja-en/ja-en -onlyaldumps 1 -p0 0.999 -s >> en-ja/corpus/en.vcb -t en-ja/corpus/ja.vcb >> died with signal 11, without coredump >> >> My tagged corpora are like this: >> head -n 1 data/tagged/train.?? >> ==> data/tagged/train.en <== >> known|VBN as|IN sesshu|NN (|-LRB- 1420|CD -|: 1506|CD )|-RRB- ,|, he|PRP >> was|VBD an|DT ink|JJ painter|NN and|CC zen|NN monk|NN active|JJ in|IN the|DT >> muromachi|JJ period|NN in|IN the|DT latter|JJ half|NN of|IN the|DT 15th|JJ >> century|NN ,|, and|CC was|VBD called|VBN a|DT master|NN painter|NN .|. >> >> ==> data/tagged/train.ja <== >> 雪舟|名詞 （|記号 せっしゅう|名詞 、|記号 年|名詞 （|記号 年|名詞 ）|記号 年|名詞 （|記号 永|名詞 正|名詞 年|名詞 ）|記号 >> ）|記号 は|助詞 号|名詞 で|助詞 、|記号 世紀|名詞 後半|名詞 室町|名詞 時代|名詞 に|助詞 活躍|名詞 し|動詞 た|助動詞 水墨|名詞 >> 画家|名詞 ・|記号 禅僧|名詞 で|助詞 、|記号 画聖|名詞 と|助詞 も|助詞 称え|動詞 られる|動詞 。|記号 >> >> Anyone know why it fails? >> >> Best regards >> -- >> Alex Helle >> >> >> _______________________________________________ >> Moses-support mailing >> [email protected]http://mailman.mit.edu/mailman/listinfo/moses-support >> >> >> >> -- >> MSc. Inf. Eleftherios Avramidis >> DFKI GmbH, Alt-Moabit 91c, 10559 Berlin >> Tel. +49-30 238 95-1806 >> >> Fax. +49-30 238 95-1810 >> >> ------------------------------------------------------------------------------------------- >> Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH >> Firmensitz: Trippstadter Strasse 122, D-67663 Kaiserslautern >> >> Geschaeftsfuehrung: >> Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender) >> Dr. Walter Olthoff >> >> Vorsitzender des Aufsichtsrats: >> Prof. Dr. h.c. Hans A. Aukes >> >> Amtsgericht Kaiserslautern, HRB 2313 >> ------------------------------------------------------------------------------------------- >> >> > > > -- > Alex Helle > > > > -- > MSc. Inf. Eleftherios Avramidis > DFKI GmbH, Alt-Moabit 91c, 10559 Berlin > Tel. +49-30 238 95-1806 > > Fax. +49-30 238 95-1810 > > ------------------------------------------------------------------------------------------- > Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH > Firmensitz: Trippstadter Strasse 122, D-67663 Kaiserslautern > > Geschaeftsfuehrung: > Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender) > Dr. Walter Olthoff > > Vorsitzender des Aufsichtsrats: > Prof. Dr. h.c. Hans A. Aukes > > Amtsgericht Kaiserslautern, HRB 2313 > ------------------------------------------------------------------------------------------- > > -- Alex Helle
_______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
