|
From: | Joseph Heled |
Subject: | Re: [Fwd: Re: [Bug-gnubg] Re: Battle of the Bots] |
Date: | Wed, 08 Jan 2003 23:53:28 +1300 |
User-agent: | Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2) Gecko/20021202 |
When I manually run game after game I get (383 - 186)/400 = 0.492 The score (after 400 games) is: gnubg 186, user 383 (money session, without Jacoby rule). So running one by one seems OK. Yet I can't find anything wrong with the rollout code. -Joseph Joseph Heled wrote:
I don't think so. When I let gnubg play at 0 ply, all seems fine. I even compared one game in fibs2html, and all is fine. Only when I do the rollout I seem to get results much different than with my own code. I have no idea what it could be.-Joseph Joern Thyssen wrote:On Mon, Jan 06, 2003 at 12:14:29PM +1300, Joseph Heled wroteIn my small sample (48 games), W won 23 time, which is 48%which is obviously without variance reduction. Could variance reduction be destroying the result? Jørn _______________________________________________ Bug-gnubg mailing list address@hidden http://mail.gnu.org/mailman/listinfo/bug-gnubg_______________________________________________ Bug-gnubg mailing list address@hidden http://mail.gnu.org/mailman/listinfo/bug-gnubg
[Prev in Thread] | Current Thread | [Next in Thread] |