Author: Mogens Larsen
Date: 06:46:09 07/24/01
Go up one level in this thread
On July 24, 2001 at 09:14:43, Bertil Eklund wrote: >Tiger Chessbase have the same learning as Deep Fritz. And of course a winning >line against A can often be a losing line against B. > >I think it should suit you if you at least pretended to be impartial when you >comment on matches and programs. > >In my last match Tiger14 won the last 8 games in a row against Hiarcs7.32, the >results in the first 30 games wasn't particularly good for Tiger, I guess that >you don't complain in this case or..... I don't think your response is sufficient considering the question raised. There are two ways of performing matches with book learning IMO. The first one is indvidual books with the weights reset before and after the match. The second one is a book for each engine that is _never_ reset. Even when switching computers. The latter option can cause some problems with several different matchmakers as in the case of the SSDF. The former option is virtually foolproof. Which one do you use? Mogens.
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.