Isn't it wonderfully misleading? You'll find the answer in "Comparison with Other Toolkits" in http://kheafield.com/code/kenlm/estimation/ . And read the rest of the document too...
Kenneth On 07/08/13 08:54, Li Xiang wrote: > I find the LM generated by SRILM lacks of many ngram when order >= 3. > with the ngram-count help command, I guess the option '-minprune' may > work. So I set minprune to 6 for forbidding pruning all single ngrams. > But in fact it does not work. > Can you tell me how to forbid pruning single ngrams? > > -- > Xiang Li > > > _______________________________________________ > 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
