Excellent. Thanks for the results. One question: do you know if this
4.12% improvement is true of the 2-ply play as well? Of course, the
0-ply results are remarkable and important, but the truth is, I analyze
everything with 2-ply and also play GNU on 2-ply alone. The only time I
use 0-ply is for rollouts, so while the 0-ply results help tell me some
of the story they don't tell me all of it. I've wondered whether by
going to 2-ply the program didn't already compensate some blindness or
imprecisions and thus the difference in strength between the 0.12 2-ply
and the 0.14 2-ply would be less. Or would it be even more as the added
improvement increases as each ply of the improved net deepens GNU's
play?
Albert
-----Original Message-----
From: address@hidden [mailto:bug-gnubg-
address@hidden On Behalf Of Joseph Heled
Sent: Friday, June 13, 2003 3:11 AM
To: Albert Silver
Cc: GNUBackgammon bug reporting
Subject: Re: [Bug-gnubg] Progress
I did a 5000,000 7ptrs between 0.14 and 0.12 - result
260294 - 239706 (52.06%)
-Joseph
Albert Silver wrote:
A poster in GoL asked:
"Do you know how much the improvement was for gnubg 0.13 as a whole,
over 0.12? That must've been even more then I suppose?"
The question means including the new race net, the crashed net, and
the
contact net.
Albert
_______________________________________________
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
_______________________________________________
Bug-gnubg mailing list
address@hidden
http://mail.gnu.org/mailman/listinfo/bug-gnubg