Hi Moses, Revisions [4347, 4330] of kenlm were sometimes refusing to build trie models from ARPAs generated by SRILM. This has been fixed. If you used the probing model or used IRSTLM to generate your ARPA, you should have been unaffected.
Kenneth On 10/08/11 11:46, Kenneth Heafield wrote: > Hi, > > This looks like a bug in the trie implementation due to some recent > changes I made for left state minimization. I'll fix it soon. A > workaround is to pass a large -m option to build_binary. > > Sorry, > > Kenneth > _______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
