[hts-users:01365] Re: Question about HERest in HTS2.1:

["Heiga ZEN (Byung Ha CHUN)" <zen@xxxxxxxxxxxxxxx>]

Hi,

QiXiong Tu wrote (2008/04/28 14:57):

> When using HERest in HTS2.1 to train the fullcontext model for the 
> second time(or more times), it's said that:
> ###########################################################
> Processing Data: 100001.cmp; Label 100001.lab
> Retrying Beta pass at 1100.0
> ...............................
> Retrying Beta pass at 5000.0
> No path found in beta pass
>  WARNING [-7324]  StepBack: Bad data or over pruning
>  in D:\LPCC\Tom\data\tools\HTS2.1RC\HEREST.exe
> ##########################################################
>  
> However,we can train the fullcontext model for several times in HTS2.0.
> Then my questions are:
> 1.Can we train the fullcontext model for several times in HTS2.1?  It's 
> necessary?
> 2.And why?

I guess you turned off variance flooring in HTS-2.1 but turned on it in HTS-2.0.

Regards,

Heiga ZEN (Byung Ha CHUN)

--
------------------------------------------------
Heiga ZEN     (in Japanese pronunciation)
Byung Ha CHUN (in Korean pronunciation)

Department of Computer Science and Engineering
Nagoya Institute of Technology
Gokiso-cho, Showa-ku, Nagoya 466-8555 Japan

http://www.sp.nitech.ac.jp/~zen
------------------------------------------------