Author: Dieter Buerssner
Date: 01:09:35 05/22/05
Go up one level in this thread
On May 21, 2005 at 17:44:23, pavel wrote: Hi Pavel, >* Yace is playing as a UCI engine under ChessbaseGUI using it's own book, I >don't know how Yace's learning function works as a UCI engine under CB GUI. It should work. I implemented it before the SSDF games. I implemented it at the same time I implemented the "autoplayer delay". From your jpg, I can see, that it is in the version you are using. >* I don't know if Yace used it's book learning in the first 1000 games, but I do >remember that at some point I added this line: "min_learn_tc 60" in Yace.ini >which you can see here: http://www.pavs.org/yace.ini . Can you try to find out at which point you added this? Perhaps the date of yace.ini. Or entries in bklearn2.pgn? >* When I run Yace under Chessbase these are the default value: >http://www.pavs.org/yace.jpg , which I never changed throughout the games. >You will see that Yace has book learning but it's book is set as random. Yes. Everything else seems default. >IMO Yace learning is clearly helping here. If you just consider the last 2-3-4 >thousands games you will probably see that Yace performance was superior. > >The question is does this learning actually help yace when playing longer time >control? ie, 15min/side? That is a good question. Perhaps, if you still have enough patience left after the match, you could try to do two 1000 game matches at 5 minutes. One with all the learning from the previus match deleted, and one with that learning. Perhaps 100 games are already enough, if the results are clearly different. Regards, Dieter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.