[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] Semeai tuning
From: |
Gunnar Farnebäck |
Subject: |
Re: [gnugo-devel] Semeai tuning |
Date: |
Mon, 17 Oct 2005 23:24:21 +0200 |
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/21.3 (sparc-sun-solaris2.9) MULE/5.0 (SAKAKI) |
Nando wrote:
> Basically, this one is just a tuning that increases the values of tactical
> attacks on important semeai worms. Additionally, attacks against defenseless,
> but important semeai worms are also added to the moves list (they were
> previously ignored). The breakage looks good :
>
> nngs:304 PASS M7 [B14|M7]
> nngs:510 PASS G14 [G14]
> semeai:60 PASS 0 0 PASS [0 0 PASS]
>
> where nngs:304 is more or less accidental (the failure in current CVS is
> only due to a cache effect), the two others are good. The impact on
> performance is positive, although negligible.
Looks good. Feel free to add it as a ticket at http://trac.gnugo.org
as well.
While we're talking about semeai, I've realized that I have forgot to
mention a fairly comprehensive semeai test suite with no less than 722
tests! It has been compiled by Ricard Vilà and can be found at
http://gobase.org/reading/preview/Semeai/#STS. Although it's in GTP
format it uses a custom command solve_semeaiS so it would be good to
adapt it to the normal GNU Go commands. For more information and a
solve_semeaiS implementation, see
http://trac.gnugo.org/gnugo/ticket/41.
/Gunnar