Thanks, James! I'll try it as you said.
About translation quality: I thought that with bigger data I'll get
more accurate translation, but now I see that translation quality got
worse than that of with less data.

Sehrob

On 8/19/14, Read, James C <[email protected]> wrote:
> add 2> output-file to your command and your STDERR messages will be dumped
> to this file instead of output to STDOUT.
>
> I have no idea why you are not getting the results you are expecting. I
> expect you will have to give more information.
>
> James
>
> ________________________________________
> From: [email protected] [[email protected]] on
> behalf of Sehrob Ibrohimov [[email protected]]
> Sent: 19 August 2014 05:47
> To: moses-support
> Subject: [Moses-support] translation quality
>
> Hi everybody !
>
> I have trained a translation model with moses (with mgiza) but not
> satisfied with the quality of translation. there are about 137000
> sentence pairs in my corpora, mainly legal acts. I had trained a
> translation model with moses before (with GIZA++) with less sentence
> pairs, about 50000, but translation was better than this. I guess that
> I have some problems with my corpora, but can't find out what exactly
> it is.
> And also while training I got this message, "nohup: ignoring input and
> redirecting stderr to stdout"
> this was my command:
> "nohup nice ~/mosesdecoder/scripts/training/train-model.perl -root-dir
> train  -corpus ~/corpus/my_corpus  -f fr -e en -alignment
> grow-diag-final-and -reordering msd-bidirectional-fe -lm
> 0:6:$HOME/lm/my_language_model:8 -mgiza  -external-bin-dir
> ~/mosesdecoder/tools > training.out "
> Could you, please, tell me if you have any idea about this problem?
>
> thanks in advance !
>
> Sehrob
> _______________________________________________
> 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