Well, ProcessPhraseTable did work with your suggested syntax/calling sequence. The collection of binary files produced is about 1Gb (compared to the 2.3Gb un-binarized phrase table), so that's some progress.
BUT... The loading of the model still grows memory usage by more than 2Gb until it reaches the Windows limit and fails. This despite the claim that: > Therefore, moses supports a binary phrase table with on-demand > loading, i.e. only the part of the phrase table that is required to > translate a sentence is loaded into memory. It looks like binarizing makes no difference to memory usage, and on-demand loading is ineffective ? <confused look> (yes, before you ask, I've changed moses.ini to point at the binary model, and for additional safety, moved the non-binary files away) -- Hubert Crépy _______________________________________________ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support