Hi, the "filter-phrase-table.perl" script that you may be referring to, removes all phrase pairs for which the source phrase does not occur in the specified test set. No probabilities are adjusted.
During training, phrases up to the length of 7 are used, and that number may be increased (and may affect inverse translation probabilities for shorter phrases). -phi On Fri, May 15, 2009 at 12:42 PM, Sanne Korzec <[email protected]> wrote: > Hi, > > > > I would like to know how the phrase table is filtered. Are phrase pairs > simply deleted when they are longer than a certain phrase length, or are the > left over probabilities also re-estimated? To sum up to one again, for > instance. > > > > Thanks, > > Sanne > > > > > > _______________________________________________ > 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
