no. if you want only some factors on the target side for chunk LM, you 
may consider an IRSTLM feature
http://article.gmane.org/gmane.comp.nlp.moses.user/4056/match=irstlm+chunk
if you need it on the source side, then some hacking is needed

On 08/04/2011 08:52, Thomas Meyer wrote:
> Hi,
>
> Is there a possibility to train a factored translation model with moses,
> when only certain words in a corpus have a factor?
>
> Or does (as with POS models) every word have to have a factor?
>
> I already tried "train-model.pl" with such a corpus, and it ends with an
> error on the first word in the corpus (which has no factor), that the
> factor 1 could not be found.
>
> Thank you very much for any hints,
> Best,
> Thomas
> _______________________________________________
> Moses-support mailing list
> [email protected]
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to