if u're comparing the speed between (cube pruning on windows) and (normal search on linux) the slowness is prob. due to the different language model implmenetations used. the windows version doesn't support SRILM or IRSTLM, instead it uses the the built-in LM which is slow & memory hungry Hieu Hoang www.hoang.co.uk/hieu
_____ From: [email protected] [mailto:[email protected]] On Behalf Of zhmmc Sent: 20 February 2009 03:30 To: [email protected] Subject: [Moses-support] Problem with the search algorithm in moses Dear expert: Now I find a problem with my new-built moses system.The moses has provided a new algorithm,cube pruning,to decode.And this new algorithm speeds up the decoding process.I set the algorithm option of the moses to cube pruning (stack size :1000)and test it under the linux OS.The result confirms that cube pruning is faster than the normal beam-search algorithm. But,when I do the same thing under the WINDOWS OS,set the algorithm to cube pruning, the complex problem appears: cube pruning works even slower than the normal beam-search algorithm. I can't find the reason.Can you help me solve the problem ? I'm eager for your answer Thanks. Best Wishes Yours zhu hai _____ 网易邮箱，中国第一大电子邮件服务商 <http://www.yeah.net>
_______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
