Author: Tim Foden
Date: 04:52:30 01/15/04
Go up one level in this thread
On January 15, 2004 at 06:35:45, Tord Romstad wrote: >My conclusion: At least against Gothmog's wide and untuned opening book, >Phalanx's learning function is extremely effective, and might be worth >more than 100 Elo points in long matches. > >Tord I would concur with this from my experience. I don't have any hard numbers, and I'm not sure it would be as much as 100 Elo, but it sure makes a detectable difference, IMO. Before I figured out tht Phalanx was using learning I used to play matches (50 games, 40/5) overnight against GLC. I would watch some of the games, and early GLC on in the match was generally doing OK, than then when I saw the result in the morning it had lost (I think this was 2.13 or so). This happened a number of times, and has happened since with other engines before I figured out that they were learning. Cheers, Tim.
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.