Hi,

a common thing to do is to break up the training data,
train separate models for each set and interpolate them.

-phi

On Wed, Dec 3, 2008 at 12:44 PM, Ibrahim Ahmed Mohammady
<[EMAIL PROTECTED]>wrote:

>  Hi All,
>
>
>
> I have a Problem with the SRI Toolkit that  when I try to generate a
>  Language Model  of size  greater than   1 Giga Words , the SRI fails
> because of the memory limit.
>
> The computer I use has two CPU Quad Core  with 16 Giga Ram.
>
>
>
> I tried the IRSTLM but it gives lower accuracy , So I prefer to use the SRI
> tool Kit .
>
>
>
> Is there any Version of SRI that overcomes  this problem or any solution
> for that problem ?
>
>
>
> Thanks ,
>
>
>
> *Ibrahim Ahmed  Mohammedy *
>
> *Senior Software Developer*
>
> *Sakhr Company*
>
> [image: cid:[email protected]][image:
> GoldCertifiedPartnerLogo]
>
>
>
> _______________________________________________
> Moses-support mailing list
> [email protected]
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>

<<image002.gif>>

<<image001.jpg>>

_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to