Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: F7 no learning - how weird

Author: Mogens Larsen

Date: 10:59:51 01/26/02

Go up one level in this thread


On January 26, 2002 at 13:35:14, karen Dall Lynn wrote:

>Yes, we have something here.
>
>I am starting to work with the hypothesis that Fritz 7 evaluates its own book
>moves in a much more strict manner than Fritz 6 - which quicly changes the
>weights. When I made the tests I played fast bullet against Fritz with random
>moves. And resigned. It is not impossible that the whole string of moves may
>have looked to Fritz as rubbish and rubbish it surely was.

:-))

I don't think Fritz have sufficient AI to make a distinction between winning
against someone playing rubbish moves and someone losing the old fashioned way
:-).

>However it is still obscure for me why Fritz does not change the evaluation of
>the white moves when I play black against it and lose, with ordinary average
>player's black moves in a non-rated game. In this case there is no rubbish yet
>the weights remain unrevised.

It is possible that learning is restricted to rated games when it comes to Fritz
7. With Nimzo 8 learning seem to be working without insisting on a rated game,
eg. a normal blitz game. The only other thing I can think of would be the book
options, where the learning strength might be set too low. I'm certain that
you've checked that already.

>Thank you for your reply.

No problem. I'm sorry that I can't be of more help. But I don't own Fritz 7.

Good luck,
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.