[Moses-support] tuning run

Hello there, Quick Qn: it is my first time to tune the decoder, I use 2500 parallel sentences for tunning and run1.mert.log has 0.200397 and, after taking about 7 hours, the 5th run5.mert.log has 0.224502 -How many times moses mert should run for the better weights, shoold I wait further? -is i

[Moses-support] distortion probablity

hello Moses Team, Quick Qns: -is there other mechanisms in phrase based SMT, that can perform better than distortion probability in reordering of target language? -can Moses work with tagged and morphological analyzed data? and is it more advantageous? Thank you. Yared. _

[Moses-support] how to put the system on the web

Hello Moses Team; I have been working on building translation system for my thesis, and it has some reordering problems still. but I want to put it on the web to translate given sentences like the moses demo(translate an input sentence to the target) and there is no much resources on how. please i

[Moses-support] how to use pre-ordering

Hello moses Team, any one who have used one sided, English parser like enju and and use head finalization rule to restructure the english order to form SOV structure. please guide me through. thank you. Yared. ___ Moses-support mailing list Moses-suppor

[Moses-support] Hierarchical model training

Hello Everyone,I posted this message before and I did not get replay, any quick suggestions pls help, I read the page http://www.statmt.org/moses/?n=Moses.SyntaxTutorial and I try to train the model using the command by just adding the following parameters from the one used to train phrase based m

[Moses-support] Hierarchical model

Hello Moses Team, I read the page http://www.statmt.org/moses/?n=Moses.SyntaxTutorial and I try to train the model using the following command: nohup nice moses/moses-scripts/scripts-20120409-0748/training/train-model.perl -scripts-root-dir moses/moses-scripts/scripts-20110409-0748/ -root-dir work

[Moses-support] which approach?

we know that english has "subject verb object" the language I want to translate english to,(Amharic) has "subject object verb" structure. - which approach phrase based, hierarchical or syntax model is preferred in this case? - and what is the criteria to select one of these approaches? any sugge

[Moses-support] Alignment and Reordering

Hello Moses Team, as we all know english has "subject verb object" the language I want to translate english to,(Amharic) has "subject object verb" structure. my question is how giza align words from the corpus? and is it necessary to use supervised alignment like Barkley tool. and I don't have c

[Moses-support] Score with NIST-BLEU

On 4/29/12, Yared Mekuria wrote: Hi Daniel, I use mteval-v13a.pl from recent generic folder of the released scripts, https://github.com/srush/transforest/blob/297ed0b0f473b09556912c9c1059468389ed02e4/mteval-v13a.pl and I use the command: mose/mteval-v13a.pl -s data/devtest/nc-test2007

[Moses-support] NIST scoring tool

Hi Daniel, thank you for your replay, I make the cased data the tokenized English corpus "news-commentary.tok.en" file, since the lower cased data was "news-commentary.lowercased.en" and it works as you say. - am I right to use the tokenized cased data? - and I can't get a NIST scoring tool.

Re: [Moses-support] Moses-support Digest, Vol 66, Issue 58

; > > Today's Topics: > > 1. how to train the recaser (Yared Mekuria) > > > ---------- > > Message: 1 > Date: Mon, 23 Apr 2012 10:24:33 -0400 > From: Yared Mekuria > Subject: [Moses-support] how to t

[Moses-support] how to train the recaser

how to train the recaser I use the following commands mkdir work/recaser /home/admin1/mose/moses-scripts/scripts-20120409-0748/recaser/train-recaser.perl -train-script /home/admin1/mose/moses-scripts/scripts-20120409-0748/training/train-model.perl -ngram-count mose/bin/irstlm/bin/i686/ngram-cou

[Moses-support] Tuning

Hello There, I am on tuning part, but what is the difference of the tunning data from training data? and while I am trying to tune using command on http://www.statmt.org/moses_steps.html nohup nice mose/moses-scripts/scripts-20120409-0748/training/mert-moses.pl worked/tuning/nc-dev2007.lowercased.

Re: [Moses-support] checking trained model

library ERROR:no LM created. We probably don't have it compiled On 4/22/12, Yared Mekuria wrote: > the step to train the model end up with the moses.ini, phrase-table, > reordering-table. > > I use the command below to check trained model > > echo "c' est une petite

[Moses-support] checking trained model

the step to train the model end up with the moses.ini, phrase-table, reordering-table. I use the command below to check trained model echo "c' est une petite maison ." | TMP=/tmp mose/moses/moses-cmd/src/moses -f works/model/moses-bin.ini Defined parameters (per moses.ini or switch): con

[Moses-support] Training Phrase Model

Hello there, I am using sentences in Welsh and English to train. I tokenized, lower cased and build language model but when I try to train Phrase model using the command below nohup nice mose/moses-scripts/scripts-20120409-0748/training/train-factored-phrase-model.perl -scripts-root-dir mose/moses

[Moses-support] srilm/Irstlm

what are the differences between the Srilm to Irstlm toolkit? which toolkit is recommended or better(srilm/Irstlm)? ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support

[Moses-support] using IRSTLM to build a tri-gram language model.

hello there, I am using welsh and english parallel corpus to test the moses translation. I tokenize and lower case the training data and while I am using IRSTLM to build a tri-gram language model using the command export IRSTLM=/home/guest/tools/irstlmtools/irstlm/bin/build-lm.sh -t /tmp -i work/l

[Moses-support] LOOKING FOR SUPPORT

Hello Moses support @mit I had hard time installing moses. and at last while I try to test the installation with sample model I get the following error. Loading lexical distortion models...have 0 models Start loading LanguageModel lm/europarl.srilm.gz : [0.000] seconds util/file.cc:33 in int util: