[Moses-support] Reference about tuning the BLEU coefficients

Dear list, the correlation of BLEU with manual measurements of quality has extensively been studied (Callison-Burch et al., WMT "Findings" paper). But, would anyone know of any paper where people have actually tuned the BLEU coefficients to approximate some kind of manual measurement of qualit

[Moses-support] Using factor in the hierarchical model

I want to use some factor (class) in the hierarchical model. I am using the ems script and using KenLM for surface and factor both. The building of LM is successful with settings "-discount_fallback" but I got an error during tuning phase like this. Exception: moses/LM/Ken.cpp:399 in Moses::Langu

[Moses-support] facored LMs in Moses

I would like to have an LM over several factors (in a factored phrase-based model). What would be the proper way to train and integrate such a model in Moses? Is it straightforward to use the factored models in SRILM? Can they be binarized in the same way as other models? Could I otherwise als

Re: [Moses-support] Learning the moses.ini v2 format

Hi Tom On 02/02/15 01:33, Tom Hoar wrote: Much of the v2 moses.ini looks self-explanatory, but I'd like to confirm my understanding. The website (http://www.statmt.org/moses/?n=Moses.FeatureFunctions) defines three feature/functions without arguments. In the moses.ini files made by train-mod

Re: [Moses-support] Using factor in the hierarchical model

Hi, It should be fixed. Whomever wrote that argument parser (not me) needs to be reminded to consume only the first = sign. Kenneth On 02/02/2015 05:06 AM, ekkim214 wrote: > > I want to use some factor (class) in the hierarchical model. > I am using the ems script and using KenLM for s

Re: [Moses-support] Learning the moses.ini v2 format

Thanks, Hieu. Here are the two scripts. Add them to the contrib (or other) folder as you see fit. They have a -h switch. The moses2-to-ini.py script copies everything from moses.ini v2 format into a traditional INI file format. The ini-2-moses2.py script restores the moses.ini v2 format. The

[Moses-support] experiment.perl: command not found

Dear All , I try to run this http://www.statmt.org/moses/?n=FactoredTraining.EMS and I attached the conflagration file after Modify. but when I run this command "experiment.perl -config config.toy" I get this error "experiment.perl: command not found". what the problem? thank you, Fatma El-Z

Re: [Moses-support] experiment.perl: command not found

Hi Fatma, experiment.perl exists in mosesdecoder/scripts/ems/. So to run with config.toy, you should say ../experiment.perl -config config.toy If everything looks okay after that. You can run ../experiment.perl -config config.toy -exec to run it. Hassan On Mon, Feb 2, 2015 at 6:23 PM, fatma elz

[Moses-support] 2nd CfP - Volume on Multi-word Units in Machine Translation and Translation Technology

Volume on Multi-word Units in MachineTranslation and Translation Technology   2 nd CALL FOR PAPERS     Background: Following the successful workshop onMulti-word Units in Machine Translation and Translation Technology held inconjunction with the MT SUMMIT 2013 conference in Nice (MUMTTT 2013

Re: [Moses-support] Reference about tuning the BLEU coefficients

Hi, while the original BLEU paper mentions co-efficients, there are not used (i.e., all are set to 1/4). There have efforts to create tunable metrics, but these are not widely used. -phi On Mon, Feb 2, 2015 at 4:34 AM, Mikel L. Forcada wrote: > Dear list, > > the correlation of BLEU with manua

Re: [Moses-support] Clarification of "Compact Phrase Table" instructions

Hi, I fixed the web site. -phi On Sun, Feb 1, 2015 at 3:17 AM, Tom Hoar wrote: > Thanks guys. > > I thought remembered the change. Website just just gave me a double-take. > > > On 02/01/2015 03:10 PM, Marcin Junczys-Dowmunt wrote: > > Hi, > yes, that should be "with 4 scores". Also "-nscores 4

Re: [Moses-support] Reference about tuning the BLEU coefficients

Thanks a million, Philipp! About these tunable metrics, has anything been published? do you have any references? Cheers Mikel > > Hi, > > while the original BLEU paper mentions co-efficients, there are not > used (i.e., all are set to 1/4). > > There have efforts to create tunable metrics, but