[computer-go] thoughts on 100,000 cgos games

[alain Baeckeroot]

Le Mardi 2 Mai 2006 02:46, Chris Fant a écrit :
> Isn't #3 fixed by only learning based on games that SlugGo loses?
> _______________________________________________

I agree with this idea. Weak bots are strong at bug squashing for strong bots.

Nealy each time GNU Go 3.7.4 loses against a much weaker bot (200+ ELO below),
there is a terrible move somewhere, and even sometimes just a bug.

Alain
 "Even the powerful have need of the weakest"
 in The Lion and the Mouse, ~600 BC, Aesop :)