Author: Christophe Theron
Date: 08:29:31 11/02/99
Go up one level in this thread
On November 02, 1999 at 02:59:37, Shep wrote:
>On November 01, 1999 at 03:44:49, Didzis Cirulis wrote:
>
>>... or 12-10 if doubles counted in.
>>
>>Well... a good result for CM 7000.
>>
>>Chess Tiger 12.0 - Fritz 6 started. Tiger leading 1.5-0.5
>
>I have just finished my 10-game series CT12 vs. F6 at 30/game (2xPIII-550),
>and Tiger won impressively 8.5:1.5.
>(Games will be posted later.)
>However, if you remove doubles, it's only 6.5:1.5 because games #7 and #9 were
>identical to #5. Seems not even book learning is working anymore for Fritz... :(
That's indeed very surprising. In fact I'm still wondering if I am right to
refuse to do agressive learning. Tiger does not try to repeat won games (the
double you and Didzis had are completely pure luck).
Maybe it's stupid to do so. I could certainly gain more points against non
learning programs (or defective learning programs :) like for example Fritz3 if
I tried to repeat won games... I have seen that the SSDF played Tiger12 against
Fritz3, in this kind of matches you get more points by repeating won games.
Maybe also agressive learning, if it fails to repeat a won game, can at least
drive the program into the kind of position it likes (or understand better)
I should maybe consider my position again.
Christophe
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.