Author: Ed Schröder
Date: 06:53:47 06/23/98
Go up one level in this thread
>Posted by blass uri on June 23, 1998 at 08:43:37: >In Reply to: Rebel's anti-GM option posted by Ed Schröder on June 23, 1998 >at 06:02:01: >On June 23, 1998 at 06:02:01, Ed Schröder wrote: >>For those who are interested (also got some questions about it) I updated >>my pages with a description and an example how Rebel will play next month >>against Anand using a special anti-GM option. >>The info is on 2 pages: >>http://www.rebel.nl >>Then from the main menu take the first 2 links from the NEWS section. >>- Ed - >I read the info and I have some questions: >The anti-GM option scores better. >can you tell us the score with the anti-GM option and without it? >what was the rate of the games(40 moves for 2 hours or faster) >did you try to do a match between Rebel10 with the anti-GM option >and Rebel10 without the anti-GM option and if the answer is yes then what >was the result. I can't provide exact figures yet. At the moment the new book for Rebel10 is tested which is pretty time consuming this in respect that a first version of a new book is always full of the usual glitches and so several updates are necessary before you again have a good and better opening book. If the Rebel10 book is okay I will place the testing results on my home page again. For the moment I can only rely on the games I have played with the Rebel9 book and the Rebel10 anti-GM option. These are simply better than without the anti-GM option. Too less games are played to justify a special page for this now. However this is of course just a matter of time. >your opinion is to set the anti-GM option OFF if I want accurate analysis >if the anti-GM option is better I do not understand why to do it. I can't tell too much at the moment, perhaps Vishy is reading CCC too :) All ok after the match but for now I must be a bit careful. However I can safely say that the goal of a Rebel engine is to give an accurate as possible main variation and score as defined in the evaluation function of the program. The anti-GM option (at least in my view) makes Rebel's returned analysis (main variation, score) and moves a bit less accurate due to the let's say "anti-GM values" which are added to the score. This with as only goal to confuse strong GM's. If you are not a GM turn the anti-GM option off :)) >I understand Rebel10 without the anti-GM option can miss good ideas >because it needs to know the idea is 100% correct and to smell the attack >is not enough for it. No that is not the case. Perhaps my description on my home page about this is a bit unclear. I will take another look at it. - Ed - >Uri
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.