Hi,

there seems to be something broken with the caching. I had
several times problems with that - maybe because just because
the features.list file was empty due to misconfiguration of an
earlier run. But this is something that could be checked against.

Also, when running Moses with -show-weights, the language and
translation models are loaded, which takes a lot of time and is really
not needed...

-phi

On Wed, Sep 28, 2011 at 10:35 AM, Hieu Hoang <[email protected]> wrote:
> Ah, it does say 'Using cached features list'
>
> you're on the money, it works now. Ta
>
> On 28 September 2011 15:40, Barry Haddow <[email protected]> wrote:
>>
>> Does the log file have this line at the start?
>>
>> Using cached features list: ./features.list
>>
>> If so, then try removing mert's tmp directory and rerun,
>>
>>
>> On Wednesday 28 September 2011 04:05:11 Hieu Hoang wrote:
>> > I just check out the latest code and ran a phrase-based model.
>> >
>> > In the 1st iteration of mert, it gives the following error :
>> >
>> > The decoder returns the scores in this order: d d d d d d d lm w tm tm
>> > tm
>> >  tm tm
>> > Mismatched lambdas. Decoder returned d d d d d d d lm w tm tm tm tm tm,
>> > we
>> > expected  at /opt/AO/sw/edinburgh-code/scripts-20110926-1425/training/
>> > mert-moses.pl line 1012.
>> > cp: cannot stat
>> >
>> > `/mnt/10.20.0.71.sdb/hieu/experiment/en-nl-Moravia/tuning/tmp.1/moses.ini':
>> > No such file or directory
>> >
>>
>> --
>> Barry Haddow
>> University of Edinburgh
>> +44 (0) 131 651 3173
>>
>> --
>> The University of Edinburgh is a charitable body, registered in
>> Scotland, with registration number SC005336.
>>
>
>
> _______________________________________________
> 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