Hi, there is currently no support for explicitly marking unknown words in the output. This is, however, something that could be added rather straightforwardly, if there are any volunteers to do it.
-phi On Tue, Jan 17, 2012 at 7:05 AM, Arda Tezcan <[email protected]> wrote: > Hi, > I was wondering if it is possible to get the unknowns marked as "UNK" on > the translation log while using mosesserver, as it seems to skip this > information (which is available with off-line decoding, with default > decoding options)? > > The difference can be seen in the two sample translation logs below (for > the same unknown word for both samples): > > *Off-line Decoding:* > Translating: unknownword > Collecting options took 0.000 seconds > Search took 0.000 seconds > BEST TRANSLATION: unknownword|UNK|UNK|UNK [1111111] ... > reset caches > > > *Mosesserver:* > Translating: unknownword > > Collecting options took 0 seconds > Search took 0 seconds > Output: unknownword > reset caches > > I tried to also get additional information by using the "verbose" option. > With the value "3" indeed I can see the unknown words being marked also on > mosesserver but this option brings out too many information (that I don't > use) and makes the process much slower when translting large files. > The best would be to just use maybe another option which displays only the > "UNK"s but any other suggestions would be very much appreciated. > > Thanks in advance, > Arda Tezcan > > > _______________________________________________ > 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
