[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

GNUbg rollouts in batch... Klaus...

Posted By: Mislav Radica
Date: Friday, 11 April 2008, at 11:44 p.m.

In Response To: GNUbg rollouts in batch... Klaus... (Klaus Evers)

Well, I think 0ply for cube is very bad and don't even consider it as relevant at all which could be enough reason. 2ply for checkers is by default better than 0ply. I don't know % of wins between these two bots but every % count. Also you are not sure in which positions 0ply agree with 2ply and in which not so I think there is no reasons with nowdays computers to use 0ply. You can't get true about position without deep rollout and this is best way and referent point for learning to know true and then have to find reasons why.

If someone play 6-8 off course it is enough for him to check with 0/0 ply evaluation. But ideal would be to have strong rollout for every position and this have to be reference point. And will be one day normal think. It is just matter of CPU power. In 2002, 2003, 2004 there were strong enough computers (Pentium III and iV) for rollouts and Snowie was there and even gnubg.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.