On Thu, Mar 24, 2011 at 10:15 AM, Lane Schwartz <dowob...@gmail.com> wrote:
> On Thu, Mar 24, 2011 at 10:07 AM, Barry Haddow <bhad...@inf.ed.ac.uk>wrote: > >> Hi Lane >> >> Ideally, moses should return the same output for the same input data. >> However, >> as Christian already noted, there are points in the code where pointer >> comparison is used to determine the order in which the search graph is >> explored, and this could potentially change the nbest lists. >> >> As far as I recall, the pointer comparison happens because we want to use >> feature state for ordering hypotheses, and for some of the language model >> features (eg SRI) the state is just a pointer to some implementation >> specific >> data. It looks to me as though kenlm implements a proper FFState object, >> as >> opposed to using PointerState, so you could try changing language model to >> see if you get deterministic behaviour. >> >> Of course, the problem you're observing could be caused by some other >> source >> of non-determinism. You should be able to observe the search graphs >> diverging >> by running with -v 3. > > > Barry & Christian, > > Thanks for the replies. FWIW, I am not using multiple threads, so that > shouldn't be an issue. > I'll give KenLM a try to see if that makes a difference. I'd been meaning > to try that out anyway. > > The pointer comparison issue seems like a likely culpruit for the case > where multiple hypotheses are tied and have different order in the n-best > lists. However, do you think that is a plausible explanation for the case > where there are translations that appear only in run A, and different > translations that only appear in run B (and the scores in these cases are > not tied)? > I'm still running more test cases, but so far using KenLM has completely eliminated nondeterministic behavior from Moses decoding. Lane
_______________________________________________ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support