Created by: zhaoningning

I have tried the chinese language Model (zh_giga.no_cna_cmn.prune01244.klm) in decoding process,but the speed is very slow,nearly 2 seconds per sample, how can i speed this?