Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty 18.12 vs 18.10 in test suites

Author: Joshua Lee

Date: 14:30:45 10/23/01

Go up one level in this thread


On October 23, 2001 at 16:36:31, Robert Hyatt wrote:

>On October 23, 2001 at 13:55:42, Joshua Lee wrote:
>
>>In my experience with Test Sets most positions are from Actual Games.
>>AEMIS I,II
>>LCT
>>ECM
>>TWGCG
>>LK      etc etc
>>
>>Nobody Should use the Test Sets not being from games as an excuse anymore.
>
>
>Depends on what the "test sets" are looking for.  Tactics is not the _only_
>part of the game of chess.  Finding tactics quicker might well mean finding
>positional stuff slower.  tactics are fine for judging tactical changes to
>the search.  But they don't help a lot when trying to estimate the playing
>strength change after a modification to the engine.




If you got my post of 1700 Positions in PGN then you should have around 400
Tactical Problems, 828 Endgame , and Nearly 500 Positional.
I would like to Reduce this more as in some cases the side to move was already
lost and the move to find didn't change a thing.... so the more of these that
are removed the better. If the positions that aren't from games can be isolated
, and the positions were the move given doesn't matter cause of a lost position
then i think it would be possible to create a better test set than the LCT II.



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.