thanks Germán
i know this is just warning, but the problem after this warning is conversion
process stoped.
the size of the phrasetable is 350MB, whereas the binary-phrasetable is 170MB?
Well... this *may* be normal... is the original phrasetable filtered
according to the test data? (i.e. using the filter-model-given-input.pl
script) I am not much of an expert at binarizing the tables myself, but I
would guess that, if the binarizer detects out-of-vocabulary words, it
might also be able to do some filtering. If nobody else answers to this
question, try using the filter-model-given-input.pl command, and then
binarize the resulting phrase table and see if that covers the size
mismatch to some extent. Note that such two commands would be the same as
using the filter-and-binarize-model-given-input.pl script, which would do
both at the same time, but we would not be able to check the sizes.
Best,
Germán Sanchis-Trilles
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
