Hi Amin,

The error message you're getting about GENERAL:order is asking you to 
specify the order of the n-gram language model. If you're using a 3-gram 
model, for instance, put "order = 3" after your "lm = ..." line.

(The reason it's giving the error as "GENERAL:order" instead of 
"LM:order" or something is that when it doesn't find it under the 
current heading, it backs off to a more general heading.)

Suzy

On 14/02/11 6:19 AM, amin farajian wrote:
>
> Hi All,
>
> I'm trying to build a Pr-En MT using moses. But since I'm using a large
> monolingual corpus (about 1.5GB), SRILM runs out of memory. I tried
> IRSTLM to build the language model and I could successfully build it
> using "build-lm.sh" and "compile-lm". But I don't know how to use it in
> EMS. When I uncomment this line in config.basic:
>
>         ### if corpus preparation should be skipped,
>
>         # point to the prepared language model
>
>         #
>
>         lm =/home/amin/mt/work/data/Pr_LM.blm
>
> running experiment.pl <http://experiment.pl/>, this error happens:
>
>         ERROR: you need to define GENERAL:order
>
> is there anyway to load the prepared language model in EMS?
> As I found from the config files, the language model must be built using
> SRILM and then IRSTLM can be used to compile it and to change its
> format(into Binary format or Quantized format). is it right? If yes, is
> there anyway to ask EMS to use IRSTLM instead of SRILM and build the
> language model using it from scratch?
>
> Regards.
>
>
>
> _______________________________________________
> Moses-support mailing list
> [email protected]
> http://mailman.mit.edu/mailman/listinfo/moses-support

-- 
Suzy Howlett
http://www.showlett.id.au/
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to