You should try to build the RandLM on the same machine as the one you're
running it on.

RandLM hasn't been tested extensively with 32-64-bit combinations, and the
main developers are long gone.

Please report back what you find.

On 16 July 2013 12:20, Gebruers Rudi <[email protected]> wrote:

>  Hi folks,****
>
> ** **
>
> I installed SRILM 1.6.0, RandLM 0.2.5, and then compiled a moses decoder
> (git version June 28) with both RandLM and SRILM on a 64-bit (openSUSE
> 12.3) platform (“build.log” in attachment).****
>
> ** **
>
> When requested to use KenLM, this moses decoder correctly loads
> “europarl.srilm.gz” (from the “sample-models” bundle) and translates “das
> ist ein kleines haus” into “this is a small house”. So far, so good.****
>
> ** **
>
> However, when I try to use the same decoder with a RandLM model that was
> generated on a 32-bit platform, I run into problems as moses simply refuses
> to load the RandLM model and dies:****
>
> ** **
>
> Defined parameters (per moses.ini or switch):****
>
>             config: x-fr.ini ****
>
>             distortion-limit: 6 ****
>
>             input-factors: 0 ****
>
>             lmodel-file: 5 0 5 /home/rgeb/smt/models/lm/x-fr.BloomMap ****
>
>             mapping: T 0 ****
>
>             ttable-file: 1 0 0 5 /home/rgeb/smt/models/pt/x-fr ****
>
>             ttable-limit: 20****
>
>             … ****
>
> *Aborted*
>
> ** **
>
> On the other hand, the very same “x-fr.BloomMap” is loaded correctly and
> used without any problem by another moses decoder which I built on a 32-bit
> (Cygwin) platform some time ago.****
>
> ** **
>
> Has anyone got a clue about what is going wrong here? Didn’t RandLM
> compile correctly (into the new moses decoder)? Am I using obsolete RandLM
> settings in my “moses.ini”? Should I somehow manipulate the RandLM model
> that was built on a 32-bit platform before I can use it on a 64-bit
> platform? …****
>
> ** **
>
> Any help would be highly appreciated.****
>
> ** **
>
> Regards,****
>
> Rudi****
>
> ** **
>
> ----- ****
>
> Translation engineer - ITP nv
> R.Gebruers @ itp-europe.com****
>
> _______________________________________________
> Moses-support mailing list
> [email protected]
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>


-- 
Hieu Hoang
Research Associate
University of Edinburgh
http://www.hoang.co.uk/hieu
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to