Re: [Moses-support] Question about corpora used for pre-trained models of release 4.0

Thanks for the quick reply! Regards Felipe El 16/4/20 a las 23:04, Hieu Hoang escribió: > they're trained on Europarl and tuned on wmt08 news test set, I believe. > > You can find the exact experimental configuration by looking at the > experiment config file, eg > http://www.statmt.org/moses/R

Re: [Moses-support] Question about corpora used for pre-trained models of release 4.0

they're trained on Europarl and tuned on wmt08 news test set, I believe. You can find the exact experimental configuration by looking at the experiment config file, eg http://www.statmt.org/moses/RELEASE-4.0/models/de-en/steps/1/config.1 Hieu Hoang http://statmt.org/hieu On Thu, 16 Apr 2

[Moses-support] Question about corpora used for pre-trained models of release 4.0

Hi all, A student of mine plans to use the phrase table available as a pre-trained model at http://www.statmt.org/moses/RELEASE-4.0/models/es-en/model/. Do you know if the test sets in WMT2006 and WMT2007 have been used for training those models. According to https://www.statmt.org/europarl/

[Moses-support] question about spliting the word

Hi, when I processed some corpus, I found many words were sticked together. e.g. 'otherdangerousmeanstosabotagefactories ' ,there are so many cases like below that I can not split it one by one，Is there any tool to handle these?Thanks very much!!! Article105.Whoever setsfires , breachesdikes ,

[Moses-support] question

Say what is the EXACT command you executed when you got the error ERROR build.log.gz Description: application/gzip ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support

[Moses-support] question in manual

Is it possible that the file "~/corpus/truecase-model.en" is created in the step in your message, so it is the output from "train-truecaser.perl"? See http://www.statmt.org/moses/?n=Moses.SupportTools#ntoc11 for details. I built the baseline system and I have that file. Support Tools - Statistic

Re: [Moses-support] question in manual

Thanks, I am all set now. On Thu, Aug 24, 2017 at 3:26 PM, Gary Hess wrote: > Is it possible that the file "*~/corpus/truecase-model.en*" is created in > the step in your message, so it is the output from "*train-truecaser.perl*"? > See http://www.statmt.org/moses/?n=Moses.SupportTools#ntoc11 fo

[Moses-support] question in manual

In section 2.3 in the example, in the truecase example, there is: ~/mosesdecoder/scripts/recaser/train-truecaser.perl \ --model ~/corpus/truecase-model.en --corpus \ ~/corpus/news-commentary-v8.fr-en.tok.en ~/mosesdecoder/scripts/recaser/train-truecaser.perl \ --model ~/corpus/truecase-model.fr --

Re: [Moses-support] Question about moses source code

Probably easiest to read the code in Moses2, a fast re-implementation of the Moses decoder. The phrase-based search is in moses2/PhraseBased/Manager.cpp The hierarchical/syntax search is in moses2/SCFG/Manager.cpp * Looking for MT/NLP opportunities * Hieu Hoang http://moses-smt.org/ On 23

[Moses-support] Question about moses source code

Sir/Ma'am, I have installed mosesdecoder and learning to use it, but i want to know which c++ modules are actually responsible for the beam search? I know that ~/mosesdecoder/search does the searching process but which what are the tasks of nbest.cc, edge_generator.cc, vertex.cc. I have read the "m

Re: [Moses-support] Question about Moses server

Is it possible to run truecaser on the client side? Its pretty straightforward to do so. If not, you can start a recasing server too ... and pass the output of translation to recaser... make sure to turn distortion-limit to 0 in recaser config :) --Prashant On Mon, Aug 22, 2016 at 2:34 PM, dai x

[Moses-support] Question about Moses server

Hi all, I am now using Moses server to do pre-translation on Across server. The problem is using Moses server, the segments are translated without tokenization and truecasing. There are huge difference between translating segments with and without tokenization and truecasing. For example, in my

Re: [Moses-support] Question about output alignment info

hi marta are there alignments in your phrase table? Can I have a look at your moses.ini file and a few lines of the input and output you're getting Hieu Hoang http://www.hoang.co.uk/hieu On 9 November 2015 at 15:10, Marta Ruiz wrote: > Hi all, > > When I use the option "-alignment-output-file

[Moses-support] Question about output alignment info

Hi all, When I use the option "-alignment-output-file [file]", I get just a few alignments. Most sentences are in blank, except some that have one alignment... best, Marta -- Marta Ruiz Costa-jussà martaruizcostaju...@gmail.com http://www.costa-jussa.com __

Re: [Moses-support] Question About matrix.stamt.org WMT 2014 Test Set

Hi, I fixed the link ;) But Matthias is right: The WMT web pages are the best place to look for links to corpora / test sets. -phi On Mon, Apr 27, 2015 at 10:56 AM, Graham Neubig wrote: > Hi Matthias, > > Thank you, that's exactly what I was looking for! > And also thanks for sending the WMT m

Re: [Moses-support] Question About matrix.stamt.org WMT 2014 Test Set

Hi Matthias, Thank you, that's exactly what I was looking for! And also thanks for sending the WMT mailing list, I'll send any further questions about the evaluation matrix to there from now on. Graham On Mon, Apr 27, 2015 at 11:31 PM, Matthias Huck wrote: > Hi Graham, > > Did you have a look

Re: [Moses-support] Question About matrix.stamt.org WMT 2014 Test Set

Hi Graham, Did you have a look at the tarballs that were distributed last year? http://www.statmt.org/wmt14/translation-task.html There are three different version: - Test sets (5.2 MB) These are the source sgm files with extra "filler" sentences. They were the actual files released for the camp

[Moses-support] Question About matrix.stamt.org WMT 2014 Test Set

Hi Moses List, Sorry about this being a bit off topic, but I have a question about the files on matrix.statmt.org, and couldn't find any information about who to contact on the site and assumed that here would be the next-best place to ask. Specifically, I'm looking for the SGM files for newstest

Re: [Moses-support] question about the transliteration from arabic to english using moses yes or no???

t the transliteration from arabic to english > using moses yes or no??? (emna hkiri) > > > ------ > > Message: 1 > Date: Mon, 6 Apr 2015 12:04:27 +0100 > From: emna hkiri > Subject: [Moses-support] question about the transliteration from >

[Moses-support] question about the transliteration from arabic to english using moses yes or no???

Dear Freinds i have a question about the transliteration from arabic to english using moses the quetion is if moses do automatically/ by default the transliteration of arabic text in the ( arabic corpus pretraitement) before he does the translation process or no thank you

[Moses-support] Question about Spanish corpora

Hi everyone, I was wondering if anybody can point to any POS-tagged, NE-tagged or lemmatised Spanish (Spain) Corpora that they used in the past and is available for academic use? Also, if anybody has some experience, I am very interested in learning about which tools are often used/preferred for

Re: [Moses-support] Question about mosesserver binary

Solution found. Previous Ubuntu versions required the package libxmlrpc-core-c3-dev. It doesn't work any more. The new package is libxmlrpc-c++8-dev. Install this package. It is currently at version 1.33 (Ubuntu 14.04.1). Then, simply include the argument --with-xmlrpc-c on the bjam command lin

Re: [Moses-support] Question about mosesserver binary

i personally compile xmlrpc-c myself and get moses to link to it ./bjam --with-xmlrpc-c=/path/to/xmlrpc-c That way, you have more control. Moses doesn't like xmlrpc-c v. 1.25.29 (off the top of my head). Also, you can delete the .so file and force moses to link to the static libraries. This w

[Moses-support] Question about mosesserver binary

I compiled the yesterday's github commit. It seems to have gone well... but I forgot to redirect output to a log file. Oops. Before restarting, I noticed the mosesserver binary is missing. On closer inspection, default Ubuntu 14.04 does not include libxmlrpc. Does anyone now the Debian package

Re: [Moses-support] question about --return-best-dev in mert-moses

OK - I see. But I can't tell you if it helps or not. I didn't do any systematic comparisons. Sorry. Thanks for the explanation. Jörg Jörg Tiedemann tiede...@gmail.com On Mar 6, 2014, at 7:37 PM, Barry Haddow wrote: > Hi Jörg > > In each MERT iteration, the first action is to decode the t

Re: [Moses-support] question about --return-best-dev in mert-moses

Hi Jörg In each MERT iteration, the first action is to decode the tuning set and create an n-best list, using the current weight set. The 1-bests from this decoding run are the hypotheses which get scored by --return-best-dev. After that decoding, MERT searchs for a weight set that can rerank t

[Moses-support] question about --return-best-dev in mert-moses

Hi, I have a question about the --return-best-dev flag in mert-moses.pl I have run several experiments using this flag and I don't really understand how it influences the choice of settings during MERT. In many cases, the system will select an early iteration which is much below in terms of BLEU

Re: [Moses-support] Question: How to parse tokens with slashes in PTB style?

Hi Martin, I don't think that will break the pipeline, but for word alignment and grammar extraction, separating out the slash character is probably a good idea for reducing data sparsity. If the reason you want PTB-style tokenization is to prepare the text for parsing then I'd recommend takin

[Moses-support] Question: How to parse tokens with slashes in PTB style?

I would like to tokenize tokens with forward slashes in the same way PTB does it. For example: Input: "Resolution 55/100" Output: "Resolution 55 / 100" (using default options) Output: "Resolution 55 %/% 100" (using "-penn" options) Desired Output: "Resolution 55/100" I skimmed through the code.

Re: [Moses-support] [Question?]How to ensure that all N-best list will be included in Moses search graph?

Hi, I've just sent you all the resources needed but i got an notification saying that my email is now being held awaiting the moderator approval (my attachment is about 13 MB). Can you approve this email or suggest me another sharing method? Thanks so much, bests, quang On Nov 4, 2013, at 1:

Re: [Moses-support] [Question?]How to ensure that all N-best list will be included in Moses search graph?

can you please send me the output search graph and the n-best list for the 1 sentence where you are seeing this error. Also, please send me the script you use to check whether a sentence is in the graph. I don't think it is possible for a sentence to be in the nbest list but not in the search gra

[Moses-support] [Question?]How to ensure that all N-best list will be included in Moses search graph?

Hi all, Currently i would like to exploit the Moses search graph as an alternative for N-best list to improve MT, thanks to a huge number of hypotheses it contains. Nevertheless when i try to output all the hypotheses in the search graph, i observe that they don't cover all N-best list. In other

Re: [Moses-support] question

Hi, the log file indicates that GIZA++ training suddenly stops. Do you have any computing restrictions on the machine you are using? Progress seems reasonable up to this point. Note that the training processes re-uses some existing files, so you want to double-check that these look okay. -phi

[Moses-support] question

Hello, I'm currently following the baseline system page (http://www.statmt.org/moses/?n=Moses.Baseline) and currently at the Training the Translation system part.I executed the command as in the tutorial, but the process finished instantly instead of supposed 1~2 hours,and instead of workin

Re: [Moses-support] Question about using multiple translation models

Hi, the alternate weight setting functionality is now documented here: http://www.statmt.org/moses/?n=Moses.AdvancedFeatures#alternateweightsettings -phi On Thu, Jun 13, 2013 at 4:11 AM, Raphael Shu wrote: > Hi, > > Really thanks for the prompt reply. > > I wonder is there any examples of how

Re: [Moses-support] Question about using multiple translation models

Hi, Really thanks for the prompt reply. I wonder is there any examples of how to using decoding graphs in moses.ini? Thanks! Raphael Shu On Thu, Jun 13, 2013 at 9:51 AM, Philipp Koehn wrote: > Hi, > > there is a replacement: alternative weight settings with the options of > turning of f

Re: [Moses-support] Question about using multiple translation models

Hi, there is a replacement: alternative weight settings with the options of turning of feature functions and decoding graphs. This still needs to be properly documented. -phi On Thu, Jun 13, 2013 at 1:44 AM, Raphael Shu wrote: > Hello all, > > I'm now testing multiple translation models in one

[Moses-support] Question about using multiple translation models

Hello all, I'm now testing multiple translation models in one moses server. (Actually I want those translation models could be accessed by one port) and I saw the class "TranslationSystem" has been removed from Moses, I tried old branches of Moses to utilize two models in one moses.ini by def

Re: [Moses-support] Question about rule table probabilities and counts

On 20/05/2013 20:39, j0hn wrote: Hello everyone, I'm trying to understand the format on the rule table from the tree based models and I'm a little bit confused about the counts. I use the following corpus to train: *Spanish:* en mi casa esta es mi casa *English:* in my house this is my hous

[Moses-support] Question about rule table probabilities and counts

Hello everyone, I'm trying to understand the format on the rule table from the tree based models and I'm a little bit confused about the counts. I use the following corpus to train: *Spanish:* en mi casa esta es mi casa *English:* in my house this is my house Then I use freeling to get the tree

Re: [Moses-support] Question

Hi Benyamin, when you're running the script filter-model-given-input.pl with this command line, /home/benyamin/mosesdecoder/scripts/training/filter-model-given-input.pl ./filtered /media/Data/translation/working/train/model/moses.ini /media/Data/translation/corpus/euro_es_en.true.es what error d

[Moses-support] Question

Hi, I'm following this tutorial : http://www.statmt.org/moses/?n=Moses.Baseline To do the last part of Tuning step, according to my personal directory (  benyamin@benyamin-desktop:/media/Data/translation/working$ nohup nice ~/mosesdecoder/scripts/training/mert-moses.pl /media/Data/translation/cor

Re: [Moses-support] Question

Hi Benyamin, to set enviroment variable IRSTLM, you have to call "export IRSLTLM=/home/benyamin/irstlm" before any other command that requires this variable to be set. Try to run this sequence instead of your single command: export IRSLTLM=/home/benyamin/irstlm irstlm/bin/add-start-end.sh < /

[Moses-support] Question

Hi, Tu run this command : irstlm/bin/add-start-end.sh < /media/Data/translation/corpus/europarl.true.en > europarl.sb.en export IRSLTLM=/home/benyamin/irstlm; irstlm/bin/build-lm.sh -i europarl.sb.en -t ./tmp -p -s improved-kneser-ney -o europarl.lm.en I have this problem : Set IRSTLM environm

Re: [Moses-support] Question

Hi Benyamin That tutorial is a wee bit out of date. Try the one on the Moses website http://www.statmt.org/moses/?n=Moses.Baseline cheers - Barry Quoting Benyamin Bosari on Mon, 15 Apr 2013 03:28:19 -0700 (PDT): > Hello all > > I am following the tutorial of moses > via http://www.statmt.

[Moses-support] Question

Hello all I am following the tutorial of moses via  http://www.statmt.org/wmt10/baseline.html To run system of development test set according to my personal directory (mosesdecoder/scripts/training/mert-moses.pl /media/Data/translation/tuning/input /media/Data/translation/tuning/reference mose

Re: [Moses-support] Question about the format of search graphs generated by moses-chart

Thank you Hieu. 2013/1/24 Hieu Hoang > The same function in ChartManager should be what you want > > Hieu > Sent while bumping into things > > On 24 Jan 2013, at 05:55 PM, "Jesús González Rubio" < > jegonza...@dsic.upv.es> wrote: > > Thanks Christian. > > I have read the code of OutputSearchNod

Re: [Moses-support] Question about the format of search graphs generated by moses-chart

The same function in ChartManager should be what you want Hieu Sent while bumping into things On 24 Jan 2013, at 05:55 PM, "Jesús González Rubio" wrote: Thanks Christian. I have read the code of OutputSearchNode and it seems to be designed to write a word-grap, not a hypergraph. ¿May be possib

Re: [Moses-support] Question about the format of search graphs generated by moses-chart

Thanks Christian. I have read the code of OutputSearchNode and it seems to be designed to write a word-grap, not a hypergraph. ¿May be possible that OutputSearchNode is the function called when the -osg option is passed to moses, and a different function is called for the same option of moses-char

Re: [Moses-support] Question about the format of search graphs generated by moses-chart

Hi, I am not aware of updated documentation on this. Your best chance is probably to read through void OutputSearchNode in moses/src/Manager.cpp which is pretty readable. cheers, Christian On 24/01/13 17:24, Jesús González Rubio wrote: > Hi, > > I'm generating some translations using the -osg

[Moses-support] Question about the format of search graphs generated by moses-chart

Hi, I'm generating some translations using the -osg option of moses-chart and I have some difficulties to fully understand the format in which the search hypergraph is outputted. ¿Is there a description of the osg format available? Cheers. -- Jesús ___

Re: [Moses-support] Question about decoding

Hello, Okay, it's clear now . Thank you Philipp 2013/1/14 Philipp Koehn > Hi, > > it is not possible that Moses gets the right translation for the word, > but the right translation is nowhere in the phrase translation table. > Unfortunately, we have not yet implemented magic, although it > wou

Re: [Moses-support] Question about decoding

Hi, it is not possible that Moses gets the right translation for the word, but the right translation is nowhere in the phrase translation table. Unfortunately, we have not yet implemented magic, although it would be a good undergraduate project. Please take a close look at the translation table.

Re: [Moses-support] Question about decoding

Hi, The problem is that the decoder traslates this sentence " " j espère qu 'il change maintenant" into its correct translation " i hope that it changes now" knowing that the single word "maintenant' does not exist in the phrase table. so my question how Moses knows that "now" is the best translat

Re: [Moses-support] Question about decoding

Hi Prashant, thank you for your answer, How the phrase is translates verbatim? any explanation please? Thnks Bests cyrine 2013/1/13 Prashant Mathur > Prashant -- *Cyrine NASRI Ph.D. Student in Computer Science* ___ Moses-support mailing list

[Moses-support] Question about decoding

Hello all, I have a question concern decoding : I run "mert" to translate this frensh sentences: " j espère qu 'il change manitenant' Moses translates it as : "i hope that it changes now" In the phrase table there are these phrases : j espère ||| i hope ||| . qu 'il ||| that it ||| But

Re: [Moses-support] Question about RPC call to Moses Server

Yes. Fong Po Po wrote: >Dear all: >  >  I have read the attached PDF file: > Barry, H. (2010). "Adding Multi-threaded Decoding to Moses." The >Prague Bulletin of Mathematical Linguistics 93: 57–66. >  On the 63th page of this file, I see these sentences: >   " M

Re: [Moses-support] Question on tuning the system with Z-mert

Hi Barry, I did it, by hacking the z-mert code and running it without using zmert-moses.pl. I also solve the big data by doing that. Hence, everything's fine. Thanks and best regards, C. Hoang On Wed, Nov 21, 2012 at 12:40 AM, Barry Haddow wrote: > Hi Cuong > > You'd be better off working with t

Re: [Moses-support] Question on tuning the system with Z-mert

Hi Cuong You'd be better off working with the Moses implementation of mert, as opposed to z-mert. It supports custom metrics. zmert-moses.pl is not really supported any more, cheers - Barry On 20/11/12 15:39, Cuong Hoang wrote: > Hi all, > > I have a problem with do experiment with my tuning m

[Moses-support] Question on tuning the system with Z-mert

Hi all, I have a problem with do experiment with my tuning method evaluation with Z-mert. I implement one evaluation metric and use it to tune the log-linear model (as suggested by Och, 2003). However, Z-mert doesn't recognize the binary-mode. When I switch the phrase-table and reodering table to

Re: [Moses-support] Question about moses chart a tree-to-* training

On 08/11/2012 12:27, Felipe Sánchez Martínez wrote: Hi Hieu, Thank you very much for your answer. Or could labels overlap in a way which is not compatible with a single tree structures? For instance: la bruja verde yep. During decoding, it will either of the incompatible labels, but no

Re: [Moses-support] Question about moses chart a tree-to-* training

Hi Hieu, Thank you very much for your answer. >> Or could labels overlap in a way which is not compatible with a single >> tree structures? For instance: >> >> >> >> >> >> >> la bruja verde > yep. During decoding, it will either of the incompatible labels, but not > both labels. I understan

Re: [Moses-support] Question about moses chart a tree-to-* training

hi felipe On 06/11/2012 03:34, Felipe Sánchez Martínez wrote: > Dear Moses-list members, > > We are trying to understand how flexible it is the tree-to-* learning > mechanism in Moses chart. Do source side annotations for an input > sentence of the form > > > > > the cat > > necessarily have to

[Moses-support] Question about moses chart a tree-to-* training

Dear Moses-list members, We are trying to understand how flexible it is the tree-to-* learning mechanism in Moses chart. Do source side annotations for an input sentence of the form the cat necessarily have to define a tree structure? Or could labels overlap in a way which is not compatibl

Re: [Moses-support] Question about using an external application as a translation model

hi vladislav yes, it's fairly easy to create a new phrase-table type. There's already 9 types! if you're doing it for phrase-based models, you would make a copy of the class PhraseDictionaryMemory and change it to whatever you need. you can run a function for each sentence by overriding

Re: [Moses-support] Question about discontinuous orientation types

ugust 2012 16:50 An: moses-support@mit.edu Betreff: [Moses-support] Question about discontinuous orientation types Hi all, What are the differences between discontinuous, discontinuous right and discontinuous left orientation in lexicalized RMs? I'm a bit lost after hours of skimming thro

[Moses-support] Question about discontinuous orientation types

Hi all, What are the differences between discontinuous, discontinuous right and discontinuous left orientation in lexicalized RMs? I'm a bit lost after hours of skimming through papers. Discontinuous orientation occurs if neither an alignment point to top left nor to the top right in an alignment

Re: [Moses-support] Question about Moses Demo

Hi, there are various ways to set it up. We have the same translation process for language pairs that share a language model (i.e., same output language). -phi On Sat, Jun 30, 2012 at 11:34 AM, Fong Po Po wrote: > Dear all: > > I know site: >http://demo.statmt.org/ >I also know ther

[Moses-support] Question about Moses Demo

Dear all:       I know site:    http://demo.statmt.org/    I also know there are five translation pairs can be chosen in the site.    I know the demo runs in Moses Server.     Does the Moses Server runs the five translation pairs in the same configuration file - moses.ini?  Does the give translati

Re: [Moses-support] Question 1 about on-line Moses System

Hi, these are currently all phrase-based systems using KenLM. -phi On Mon, May 21, 2012 at 5:25 PM, Fong Po Po wrote: > Dear all: > >I have see > http://demo.statmt.org/ > Does Mosesserver in the demo system use irstlm or srilm? > Whether Mosesserver in the demo system

[Moses-support] Question 1 about on-line Moses System

Dear all:      I have see     http://demo.statmt.org/   Does Mosesserver in the demo system use irstlm or srilm?   Whether Mosesserver in the demo system is Phrase-Based SMT or Syntax-Based SMT?   Can you help me? Thanks!   Best Regards, Fong Pui Chi___

Re: [Moses-support] Question 1 about daemon.pl

Hi Fong The Moses demo uses moses server, and doesn't use daemon.pl. I'm not familiar with daemon.pl, cheers - Barry On Monday 21 May 2012 01:41:05 Fong Po Po wrote: > Dear Dr.Barry: > > I know > http://demo.statmt.org/ > Does this demo use daemon.pl? > If this demo us

[Moses-support] Question 1 about daemon.pl

Dear Dr.Barry:     I know   http://demo.statmt.org/   Does this demo use daemon.pl?   If this demo uses daemon.pl, does it use Moses or Moses Server?   If it use daemon.pl connection ot Moses, why it uses Moses instead of Moses Server?   If I develop one Moses system using

Re: [Moses-support] Question 1 about extract.o

Hi, please check http://www.statmt.org/moses/?n=FactoredTraining.BuildReorderingModel and the papers referenced in there, especially http://homepages.inf.ed.ac.uk/pkoehn/publications/iwslt05-report.pdf In short, a line like > and products are ||| 企業界 ||| mono other indicates that the phrase pair

[Moses-support] Question 1 about extract.o

Dear All: We can get extract.o after training Moses. This extract.o like this: and products are ||| 企業界 ||| mono other losing ||| 每年 損失 ||| other mono losing millions ||| 每年 損失 數百萬 ||| other other . Is this file get from computing the probability of different types of orientation fo

[Moses-support] Question 1 about extract.o

Dear All: We can get extract.o after training Moses. This extract.o like this: and products are ||| 企業界 ||| mono other losing ||| 每年 損失 ||| other mono losing millions ||| 每年 損失 數百萬 ||| other other . Is this file get from computing the probability of different types of orientation fo

[Moses-support] Question regarding Berkeley Aligner

> > > Hello all > > I have been working with Berkeley Aligner for a while now. I had a > question > regarding it's configuration file. I am trying to align English-Hindi > parallel corpus as provided > at the following link. http://www.cse.unt.edu/~rada/wpt05/#def . This > link provided me wit

回复： [Moses-support] Question 2 in reordering table file

1801 发件人： 亮 田 收件人： Philipp Koehn 抄送： "moses-support@mit.edu" 发送日期： 2012年4月15日, 星期日, 下午 5:09 主题: 回复： [Moses-support] Question 2 in reordering table file Dear Fong Puichi,   As Philipp had said, your first file is a word alignment file ( the file name shoul be

回复： [Moses-support] Question 2 in reordering table file

收件人： Fong Po Po 抄送： Moses 发送日期： 2012年4月13日, 星期五, 上午 11:16 主题: Re: [Moses-support] Question 2 in reordering table file Hi,the first file looks like a word alignment file.Also: you are using train-factored-phrase-model.perl which indicates that you areusing a very old version of Moses. Please

Re: [Moses-support] Question 2 in reordering table file

Hi, the first file looks like a word alignment file. Also: you are using train-factored-phrase-model.perl which indicates that you are using a very old version of Moses. Please check out the latest release from the git repository. -phi On Thu, Apr 12, 2012 at 5:51 PM, Fong Po Po wrote: > Dear

[Moses-support] Question 3 in reordering table file

Dear all:      Does  Distance-based  reordering model measure reordering distance? or compute all probabilities of phrase reordering in one sentence? Thanks!   Best Regards, Fong Pui Chi___ Moses-support mailing list Moses-support@mit.edu

[Moses-support] Question 2 in reordering table file

Dear all:      When I type this command to run Moses:    /train-factored-phrase- model.perl --scripts-root-dir $SCRIPTS_ROOTDIR --corpus /home/tianliang/mosesdecoder/corpus/clean --f eng --e chn --alignment grow-diag-final-and  -reordering msd-bidirectional-fe –lm 0:5:/home/tianliang/mo

Re: [Moses-support] Question about glue-rule for hierarchical model

Yes, you're right, sorry if that was unclear. I omitted the source LHS symbol because it's always X in hierarchical phrase-based models. Phil On 10 Apr 2012, at 13:55, Bill_Lang(Gmail) wrote: > Hi Philip, > Thanks for your kindly and professional replying. I think your > answers are c

Re: [Moses-support] Question about glue-rule for hierarchical model

Hi Philip, Thanks for your kindly and professional replying. I think your answers are correct. But I have one point not totally clearly. For rule c:[X][S] [X] ||| [X][S] [S] ||| 1 ||| 0-0 ||| 0 In the left side, does the first [X][S] mean the source side target must be "X" and the ta

Re: [Moses-support] Question about glue-rule for hierarchical model

Hi, the Moses rule table format pairs up source and target non-terminal symbols for the rule right-hand-sides, so you should read the SCFGs in rule c and d as being S -> X , S and S -> X X, S X (With monotone non-terminal correspondences, i.e. no reordering.) The next steps in your de

[Moses-support] Question about glue-rule for hierarchical model

Dear Moses community friends, For moses hierarchical decoding, I have a question about the auto generated gleu-rule file. I make this example as follows. -- rule-table: aa [X] ||| bb [X] ||| 0.5 0.572507 0.000191388 1

Re: [Moses-support] question about table weights

Thanks a lot! Panos On Friday 23 of March 2012 16:22:23 Philipp Koehn wrote: Hi, this paper: Word-Based Alignment, Phrase-Based Translation: What's the Link? With Philip Resnik. In Proceedings of AMTA, pages 90–99, August 2006 has a nice discussion about which features can be turned off at w

Re: [Moses-support] question about table weights

Hi, this paper: Word-Based Alignment, Phrase-Based Translation: What's the Link?With Philip Resnik. In *Proceedings of AMTA* , pages 90–99, August 2006 has a nice discussion about which features can be turned

[Moses-support] question about table weights

Hi all, According to Moses manual, the 5 scores of each translation table are the following: 1. inverse phrase translation probability 2. inverse lexical weighting 3. direct phrase translation probability 4. direct lexical weighting 5. phrase penalty If I wanted to experiment manually with thes

Re: [Moses-support] question about changing BLEU

Le 14/02/2012 17:17, bingyuanmy a écrit : > Dear all, > > I'm using MOSES to do my task for which I have my own metrics. > So my question is how I can change the BLEU score into my metrics > to train the system. Could you give me some hints or some references. > Thanks a lot ! > > Best wishes, > Yu

Re: [Moses-support] question about changing BLEU

Hi Yue This is certainly possible, although not well documented. If you look in the mert directory of moses, you'll see that several scorers are implemented, for bleu, cder, ter and per. They should give you some indication as to how to proceed. Also, these papers may help: http://ufal.mff.cun

[Moses-support] question about changing BLEU

Dear all, I'm using MOSES to do my task for which I have my own metrics. So my question is how I can change the BLEU score into my metrics to train the system. Could you give me some hints or some references. Thanks a lot ! Best wishes, Yue ___ Moses-su

Re: [Moses-support] question about BLEU

Hi, this tool was developed by NIST, so it would be best to ask someone from there, for instance Mark Przybocki. -phi On Tue, Dec 6, 2011 at 7:08 AM, bingyuanmy wrote: > Dear all, > > By the mteval-v11b.pl script, we can get results about > "Cumulative N-gram scoring" and "Individual N-gram sco

[Moses-support] question about BLEU

Dear all, By the mteval-v11b.pl script, we can get results about "Cumulative N-gram scoring" and "Individual N-gram scoring". I don't know the differences between them, so may someone explain it to me or point me a reference that I can refer to? Thanks a lot! Best wishes, Yue

Re: [Moses-support] question about phrase-table

Hi Yue This is covered in the faq http://www.statmt.org/moses/?n=Moses.FAQ#ntoc7 For the last two numbers, see this recent mail on the list http://www.mail-archive.com/moses-support@mit.edu/msg04812.html cheers - Barry On Tuesday 22 November 2011 15:54:14 bingyuanmy wrote: > Dear all, > > I ru

[Moses-support] question about phrase-table

Dear all, I run Moses on my corpus and got the phrase-table containing information as follows: an atmosphere ||| an xxatmosphere ||| 1 0.9 1 0.9 2.718 ||| ||| 1 1 Could you kindly explain what each number means in it ? Thanks a lot! Best regards, Yue

[Moses-support] question about bleu

Hi, With help of Moses, I have built a English-Persian SMT. In comparison with Google translate, my BLEU score is less than the Google translator whereas P1, P2,P3 and P4 of my system is more than the Google translate. BP (brevity penalty) of Goolge translate is always 1 whereas my BP is .6 or .7.

Re: [Moses-support] Question on model for the demo system

Thank you so much. I will try it. Cheers, Bill Tang On Sun, Oct 2, 2011 at 12:58 PM, Philipp Koehn wrote: > Hi, > > the French-English system was trained from data available > for the WMT 2011 workshop ( http://www.statmt.org/ ). Fairly > standard settings were used. In fact, if you check out

Re: [Moses-support] Question on model for the demo system

Hi, the French-English system was trained from data available for the WMT 2011 workshop ( http://www.statmt.org/ ). Fairly standard settings were used. In fact, if you check out the example configuration files for experiment.perl, then you should be able to pretty much reproduce the model. -phi

[Moses-support] Question on model for the demo system

I tried the moses online demo and found it to be very good translating French to English. Is the model for the demo available online? I have compiled mosesdecoder, irstlm, srilm. But it appears my local build can only translate simple sentences. I am new to moses. Which language model does the o

  1   2   >