Hello, It looks like most of these games are being won in the opening. Doesn't
mogo have a big UCT opening book? Is it learning from each game it plays as well?
unfortunately no for both. Its opening book is at maximum 4 ply (deepest variation) and if you play first on a yoshi it is 2 ply (that is white answer to your move and that's all). Learning from each game would be great, but not so easy to do if you don't have access to the strength of the opponent. And by the way, it would be a little cheating, as the opening book would be then be overfitting the players on cgos. For example I managed in some experiments to create on opening book which gives a 100% winning rate against gnugo! But (i) there is bad moves, and (ii) it would be an not interesting overfitting. Sylvain PS: by the way, I set "limited" version of MoGo playing on cgos, with extension "_xk", which means making x thousands of simulations per move. Currently, there are 3k and 10k playing. David
> -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey > Sent: Friday, January 12, 2007 6:41 PM > To: computer-go > Subject: [computer-go] reign of terror > > > Someone needs to get their bot on CGOS and end Mogo's reign of terror. > > A version of MoGo has achieved a CGOS rating of well over 2300! > > > > - Don > > > _______________________________________________ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
