Author: Robert Hyatt
Date: 10:57:19 10/06/99
Go up one level in this thread
On October 06, 1999 at 11:53:42, Georg v. Zimmermann wrote: >> >>Why play a match with learning off? why not ponder=off? or with the passed >>pawn scoring turned off? or with the time allocation code turned off? An >>engine is the sum of its parts. When you turn off one part, what does that do >>to the validity of the test? > >You play program1 against p2 , then >p1 vs. p3 (same starting positions), then >p1 vs. p4 >p1 vs. p5 > >get what i mean ? Learning will help p1 while it doesn't help the other >programs, if you really like testing p1 it will get better and better :-). > >Also, he was trying to say something about whether results can reproduced. For >that he has to have the same conditions again of course. > >Obviously to say "20 games are enough, i don't care about statistic-theory" >shows that you are kind of , uhm, "special". \ I don't see that problem. That is the _reason_ I developed learning in the first place for Crafty. :) Hey... I don't have a particularly difficult to code piece of knowledge done just yet... can we turn that off in the other program? :) Bob
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.