the little
idiosyncracies, is also a very large piece of work and has an extremely large
impact on the perceived accuracy of the whole system in the real world.
(*) unintentional pun, please forgive me ;-)
--
Hubert Crépy
___
Moses-support mailing list
Philipp Koehn [EMAIL PROTECTED] writes:
for some reason that command needs to be called like this:
LC_ALL sort phrase-table | /STATMT/moses/misc/processPhraseTable
-ttable 0 0 - -nscores 5 -out phrase-table
Thanks, I'll try that !
___
Hieu Hoang [EMAIL PROTECTED] writes:
For the WMT task recasing is done by training another Moses model. Is this
a
suitable approach for the online system – would you train the recaser using
all
of the Europarl data?? Another option would be to train the original models
on
data that has
-- apologies if this is a duplicate, I'm having trouble posting to the
list --
I just built (as an exercise) a fr-en model based on the first 500K
sentences of Europarl.
It appears the generated model is just way too big to load on my Windows
machine: while trying to load, I see memory/swap
it directly on the .gz zipped table or on a previously
gunzipped version -2Gb-, makes no difference...)
--
Hubert Crépy
___
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support
In addition to my own failings and setup oddities, I found what I think
to be a small bug in the build setup:
I had to edit released-filesWIN32 to list:
training/mbr/mbr.exe
It was missing the trailing .exe and rsync was failing.
--
Hubert Crépy
-nsmooth 4 -o
working-dir/giza.fr-en/fr-en -onlyaldumps 1 -p0 0.999 -s
working-dir/corpus/en.vcb -t working-dir/corpus/fr.vcb died with signal
11, with coredump
Er, what now ? :-/
--
Hubert Crépy
___
Moses-support mailing list
Moses-support@mit.edu
-nsmooth 4 -o
working-dir/giza.fr-en/fr-en -onlyaldumps 1 -p0 0.999 -s
working-dir/corpus/en.vcb -t working-dir/corpus/fr.vcb died with signal
11, with coredump
Er, what now ? :-/
--
Hubert Crépy
___
Moses-support mailing list
Moses-support@mit.edu
support, much appreciated.
--
Hubert Crépy
___
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support