[Moses-support] baseline model run with same setting, with same database twice gives different translation output
Hi, I ran Moses baseline method twice with the same settings and with the same data set, but final translation file generated in both the cases is different. If i am doing something wrong please suggest? Thanks and regards ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
[Moses-support] How to use trained models on different system
1. My existing system is also facing some problem so I need to reinstall Ubuntu on it. Please suggest after installing Ubuntu, to install Moses should I clone Moses from https://github.com/moses-smt/mosesdecoder.git Or there is some other repository. 2. Will I be able to use previously trained models (trained earlier with some parallel data )by simply copy pasting "working" directory after installing Moses again on my system. Thanks ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
Re: [Moses-support] Moses-support Digest, Vol 153, Issue 4
Thanks a lot, Please suggest if everything is same in both runs why there is a difference in no. of files created in "mert-work" directory. I am also getting extract.err files in my mert-work directory and extract.out files are blank. this is what i get in extract.err file: Binary write mode is NOT selected Scorer type: BLEU name: case value: true Loading reference from /home/radha/Desktop/DBsets/set5/method1/corpusEH/tune.en-hi.true.hi ..Data::m_score_type BLEU Data::Scorer type from Scorer: BLEU loading nbest from run7.best100.out.gz Loaded N-best lists : [Wall 0.158054 CPU 0.156] seconds. Binary write mode is NOT selected saving the array into run7.features.dat saving the array into run7.scores.dat Stopping... : [Wall 0.38641 CPU 0.388] seconds. Also please suggest how to overcome this error and the search error. I have also checked that when I was training moses for the *wmt13 *database as given in manual(*baseline method*) then also I got these* .err *files and *extract.out *was a blank file. Please suggest if I am doing something wrong. Thanking in advance Best Regards On Sat, Jul 6, 2019 at 9:33 PM wrote: > Send Moses-support mailing list submissions to > moses-support@mit.edu > > To subscribe or unsubscribe via the World Wide Web, visit > http://mailman.mit.edu/mailman/listinfo/moses-support > or, via email, send a message with subject or body 'help' to > moses-support-requ...@mit.edu > > You can reach the person managing the list at > moses-support-ow...@mit.edu > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Moses-support digest..." > > > Today's Topics: > >1. Re: different numbers of runs of mert for samedatabase > (Philipp Koehn) > > > -- > > Message: 1 > Date: Fri, 5 Jul 2019 14:45:29 -0400 > From: Philipp Koehn > Subject: Re: [Moses-support] different numbers of runs of mert for > samedatabase > To: rmogla > Cc: "moses-support@mit.edu" > Message-ID: > qxaktez7q3...@mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > Hi, > > these numbers are computed of the merged n-best lists, so the actual BLEU > score my differ due to search error. > > -phi > > On Tue, Jul 2, 2019 at 1:46 PM rmogla wrote: > > > hello, > > I am training moses , I ran it twice with the same database using the > > baseline system but both the time mert.out file show different numbers of > > run. > > As in mert.out file: > > > > (8) BEST at 8: 0.118606 0.00894043 0.0616433 0.0270803 -0.00458744 > > 0.0636065 0.0394991 0.121296 -0.0393199 -0.349773 0.0781373 0.00882911 > > 0.0719328 0.00674891 => 0.54342 at Fri... > > > > (7) BEST at 7: 0.042261 0.0543911 0.0644467 0.0409684 -0.114557 0.0567117 > > 0.0807067 0.099067 -0.0280285 -0.277286 0.0593325 0.0104217 0.0663609 > > -0.00546069 => 0.546516 at Tue... > > > > though the BLEU score of finally trained moses varies by .01, > > > > Please suggest if I am doing something wrong or is it normal to have > > various number of mert runs for the same database. > > > > Thanking in advance. > > Best Regards > > > > > > > -- next part -- > An HTML attachment was scrubbed... > URL: > http://mailman.mit.edu/mailman/private/moses-support/attachments/20190705/79f5709b/attachment-0001.html > > -- > > ___ > Moses-support mailing list > Moses-support@mit.edu > http://mailman.mit.edu/mailman/listinfo/moses-support > > > End of Moses-support Digest, Vol 153, Issue 4 > * > ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
[Moses-support] different numbers of runs of mert for same database
hello, I am training moses , I ran it twice with the same database using the baseline system but both the time mert.out file show different numbers of run. As in mert.out file: (8) BEST at 8: 0.118606 0.00894043 0.0616433 0.0270803 -0.00458744 0.0636065 0.0394991 0.121296 -0.0393199 -0.349773 0.0781373 0.00882911 0.0719328 0.00674891 => 0.54342 at Fri... (7) BEST at 7: 0.042261 0.0543911 0.0644467 0.0409684 -0.114557 0.0567117 0.0807067 0.099067 -0.0280285 -0.277286 0.0593325 0.0104217 0.0663609 -0.00546069 => 0.546516 at Tue... though the BLEU score of finally trained moses varies by .01, Please suggest if I am doing something wrong or is it normal to have various number of mert runs for the same database. Thanking in advance. Best Regards ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
[Moses-support] Error in language model training.
Hello, While using KenLM for language model training for Hindi I got this error. mosesdecoder/lm/builder/adjust_counts.cc:61 in void lm::builder::{anonymous}::StatCollector::CalculateDiscounts(const lm::builder::DiscountConfig&) threw BadDiscountException because `discounts_[i].amount[j] < 0.0 || discounts_[i].amount[j] > j'. ERROR: 1-gram discount out of range for adjusted count 2: -0.0625 Aborted (core dumped) Please suggest how to remove this error. Thanking in advance. Best Regards ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
[Moses-support] query regarding system configuration and adding other methods to moses
Hello, I am a new user of moses, and trying to install moses and giza++. 1. Please suggest what is the minimum system configuration(RAM, memory etc.) to use moses. 2.Can we add another non statistical alignment model instead of giza++ to moses. 3. As per my understanding we can train the system on the basis of varying the phrase length and order of language model, is my understanding correct. Apart from these two things on what basis we can modify moses or is it possible and feasible to add some new methods probably a non statistical algorithm to moses? If yes in which language the codes should be written? Thanking in advance. Best Regards ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
[Moses-support] Query for moses
Hi, I am a new user of moses and using it for the first time. Ihave installed moses and giza++ on a 32 bit machine with ubuntu 15.04 , but while doing language model training and running the command ~/mosesdecoder/bin/lmplz -o 3 <~/corpus/news-commentary-v8.fr-en.true.en > news-commentary-v8.fr-en.arpa.en getting below mentioned output "Aborted (core dumped)." === 1/5 Counting and sorting n-grams === Reading /home/radha/corpus/news-commentary-v8.fr-en.true.en 5---10---15---20---25---30---35---40---45---50---55---60---65---70---75---80---85---90---95--100 tcmalloc: large alloc 2180882432 bytes == 0x35312000 @ Unigram tokens 4066699 types 62719 === 2/5 Calculating and sorting adjusted counts === Chain sizes: 1:752628 2:880841812 3:1651578397 tcmalloc: large alloc 1651580928 bytes == 0x35312000 @ in malloc *Aborted (core dumped)* I am not able to understand the reason for the error .Please suggest how to proceed further. Hoping for a quick reply and Thanking in advance Best Regards ___ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support