Quoting Michael Williams <[EMAIL PROTECTED]>:
I'm curious what the winrate of the bug-fixed version is over the
original version.
The last version with bugs was Valkyria3.2.0 with 2216 Elo on CGOS,
whereas the new version Valkyria3.2.1 currently has 2255. It is doing
really well against Mo
Magnus Persson wrote:
I have now debugged Valkyria with the test positions, 5 bugs was found
directly in the move generation preventing some tesujis to be played. 2
bugs was also found for moves deeper in the search tree. Some bugs
affected more than one position.
With the new version Valkyri
L PROTECTED] [mailto:computer-go-
> [EMAIL PROTECTED] On Behalf Of Gunnar Farnebäck
> Sent: Saturday, April 26, 2008 11:58 AM
> To: computer-go
> Subject: Re: [computer-go] Test position set for MC programs
>
> Yamato wrote:
> > I attached the fixed version to this email. Thanks for you
David Fotland wrote:
> Another alternate move. On problem 108 it seems that a2 works,
although it
> looks inferior.
> A2, a3, g1, h1, a4, and if white cuts at b4, black captures and white
has no
> ko threats.
> If white doesn’t answer at h1, black can live in the corner and white
can’t
> li
"Huh" is right. You wrote C2 C1 and so did I, but I somehow did so
while looking at C1 C2. I'm not sure how I managed that myself.
On Sat, Apr 26, 2008 at 5:24 PM, Gunnar Farnebäck <[EMAIL PROTECTED]> wrote:
> Huh? A seki doesn't get much more basic than this.
>
> 4 X X X X X . X . X 4
> 3 . O
Huh? A seki doesn't get much more basic than this.
4 X X X X X . X . X 4
3 . O O O O X X O . 3
2 O O O . O O X . . 2
1 . X X X O X X . . 1
A B C D E F G H J
/Gunnar
Eric Boesch wrote:
> Black doesn't need the lower-right corner as territory after killing
> the lower left. After B B1 W C
Black doesn't need the lower-right corner as territory after killing
the lower left. After B B1 W C2 B C1 W H3 B G3, w gains about 7
points in the lower right but has already lost like 31 in the lower
left, and black wins by 15. Either I'm missing something obvious, or
you overlooked that the lowe
correct?
David
> -Original Message-
> From: [EMAIL PROTECTED] [mailto:computer-go-
> [EMAIL PROTECTED] On Behalf Of Gunnar Farnebäck
> Sent: Tuesday, April 22, 2008 12:20 PM
> To: computer-go
> Subject: Re: [computer-go] Test position set for MC programs
>
> Y
Yamato wrote:
> I attached the fixed version to this email. Thanks for your help.
Another correction. In 119 black has a serious weakness on the right.
Is there any way for black to win after B B1, W C2, B C1, W H3?
/Gunnar
___
computer-go mailing list
I have now debugged Valkyria with the test positions, 5 bugs was found
directly in the move generation preventing some tesujis to be played.
2 bugs was also found for moves deeper in the search tree. Some bugs
affected more than one position.
With the new version Valkyria does:
100035
You are right. B2 does not work.
On Thu, Apr 24, 2008 at 11:23 PM, Yamato <[EMAIL PROTECTED]> wrote:
> John Fan wrote:
> >On Problem 125, sounds like B2 is also a right move.
>
> B:B2, W:B1, B:C1, W:A2, B:A1, W:C2, B:A3, W:D1.
>
> Black seems to cannot win this ko.
>
> --
> Yamato
> _
John Fan wrote:
>On Problem 125, sounds like B2 is also a right move.
B:B2, W:B1, B:C1, W:A2, B:A1, W:C2, B:A3, W:D1.
Black seems to cannot win this ko.
--
Yamato
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailm
When you guys get those problems corrected, can you send an updated version?
- Don
John Fan wrote:
> On Problem 125, sounds like B2 is also a right move.
>
> On Wed, Apr 23, 2008 at 3:33 PM, Gunnar Farnebäck <[EMAIL PROTECTED]>
> wrote:
>
>
>> Yamato wrote:
>>
>>> Thanks Gian-Carlo, Gunn
On Problem 125, sounds like B2 is also a right move.
On Wed, Apr 23, 2008 at 3:33 PM, Gunnar Farnebäck <[EMAIL PROTECTED]>
wrote:
> Yamato wrote:
> > Thanks Gian-Carlo, Gunnar.
> >
> > Current list of results.
> >
> > GNU Go 3.7.12 level 0 : 24/50
> > GNU Go 3.7.12 level 10 : 34/
that black is so far behind that it should politely resign :-\
- Dave Hillis
-Original Message-
From: Gunnar Farnebäck <[EMAIL PROTECTED]>
To: computer-go
Sent: Wed, 23 Apr 2008 3:33 pm
Subject: Re: [computer-go] Test position set for MC programs
Yamato wrote:
> Thanks Gian
Yamato wrote:
> Thanks Gian-Carlo, Gunnar.
>
> Current list of results.
>
> GNU Go 3.7.12 level 0 : 24/50
> GNU Go 3.7.12 level 10 : 34/50
> GNU Go 3.7.12 level 15 : 37/50
> GNU Go 3.7.12 mc, 1k : 30/50
> GNU Go 3.7.12 mc, 10k : 31/50
> GNU Go 3.7.12 mc,
Hi all, thanks Yamato for the positions.
FirstGo version 330 results:
1k playouts -> 27/50 passes10k playouts -> 30/50 passes100k playouts ->
33/50 passes
(100k takes about 8 seconds/position average on P4 2.67Ghz.)
Edward.
___
David Fotland wrote:
on problem 101, Many Faces likes A4, which I think also works, but I didn't
have time to check it thoroughly. If A4 doesn't work, then Many Faces is
back to 37.
A4 doesn't work. White plays E1 and is safe. Black G1 is answered by J3.
/Gunnar
__
I went through the problems with Valkyria.
Right now it can do 32 out 50 where it had to think more than 10
seconds on perhaps 2 of those problems it solved.
But I do recommend to go through these problems by hand. At least one
position had the first move right but it seemed very week on the
lf Of Yamato
> Sent: Wednesday, April 23, 2008 5:21 AM
> To: computer-go
> Subject: Re: [computer-go] Test position set for MC programs
>
> David Fotland wrote:
> >I didn't see this:
> >
> >>>148: D1 also wins?
> >> You are right. Thanks for correction
: Re: [computer-go] Test position set for MC programs
>
> Gunnar Farnebäck wrote:
> > >> 143: I don't see how A3 could win the semeai. A2 and C4 look more
> > >> effective.
> > >
> > > Typo, it was A2. C4 cannot work.
> >
> >How d
I tested with your old version. I'll try tomorrow with the new one
> -Original Message-
> From: [EMAIL PROTECTED] [mailto:computer-go-
> [EMAIL PROTECTED] On Behalf Of Yamato
> Sent: Wednesday, April 23, 2008 5:21 AM
> To: computer-go
> Subject: Re: [computer-go
David Fotland wrote:
>I didnt see this:
>
>>>148: D1 also wins?
>> You are right. Thanks for correction.
>
>Many Faces played D1, so change it to 38 correct.
Did you use the fixed version?
I corrected #148 as follows. Is it still wrong?
loadsgf sgf/mc148.sgf
148 reg_genmove black
#? [B1|D1]
--
nt: Tuesday, April 22, 2008 12:20 PM
> To: computer-go
> Subject: Re: [computer-go] Test position set for MC programs
>
> Yamato wrote:
> > Gunnar Farnebäck wrote:
> >>>> 143: I don't see how A3 could win the semeai. A2 and C4 look more
> >>>> effe
To: computer-go
> Subject: Re: [computer-go] Test position set for MC programs
>
> Thanks Gian-Carlo, Gunnar.
>
> Current list of results.
>
> GNU Go 3.7.12 level 0 : 24/50
> GNU Go 3.7.12 level 10 : 34/50
> GNU Go 3.7.12 level 15 : 37/50
>
Thanks Gian-Carlo, Gunnar.
Current list of results.
GNU Go 3.7.12 level 0 : 24/50
GNU Go 3.7.12 level 10 : 34/50
GNU Go 3.7.12 level 15 : 37/50
GNU Go 3.7.12 mc, 1k : 30/50
GNU Go 3.7.12 mc, 10k : 31/50
GNU Go 3.7.12 mc, 100k : 38/50
GNU Go 3.7.
Yamato wrote:
> Gunnar Farnebäck wrote:
143: I don't see how A3 could win the semeai. A2 and C4 look more
effective.
>>> Typo, it was A2. C4 cannot work.
>> How does white defend against C4? I'm looking at B C4, W B4, B B5,
>> W B6, B A2 without finding a way out for white. Did I miss so
> I attached the fixed version to this email. Thanks for your help.
Leela 0.3.14
1k -> 19/50 passes
10k -> 28/50 passes
100k -> 36/50 passes
--
GCP
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo
Yamato wrote:
> Gunnar Farnebäck wrote:
>> 130: D2 becomes more complicated but looks good enough to win with
>> komi 7.5.
>
> I don't see why D2 works. After E2, can black live?
You're right. I was busy counting how much damage white could do on
the right with various moves cutting the threatene
Gunnar Farnebäck wrote:
>108: If I'm not mistaken A3 and G1 both give B+3.5 whereas A2 gives
>B+1.5 so komi could be 9.5 but there would still be two solutions.
>
>113: Black can also start with E8 and still have sufficient ko
>threats.
>
>125: B1 also kills.
>
>128: D2 kills, not C2.
>
>131: It lo
Yamato wrote:
> Don Dailey wrote:
>> Go problems don't work for MC programs unless you can arrange them so
>> that finding the right move wins, and anything else loses. I found you
>> can modify some problems by manipulating the komi to make this true.
>
> I intend to have adjusted all of them l
Yamato wrote:
> Hi all.
>
> I'd like to share my test set for MC programs.
> This is a gnugo-style "regression test" file. To use it, you need
> to implement 2 GTP commands, loadsgf and reg_genmove.
> Then, run your program like this:
>
> gnugo --mode gtp < mctest.tst | awk -f regress.awk tst=mcte
Don Dailey wrote:
> Go problems don't work for MC programs unless you can arrange them so
> that finding the right move wins, and anything else loses. I found you
> can modify some problems by manipulating the komi to make this true.
You can manipulate komi to turn any unfinished position into
By the way, thank you for contributing this set of go problems, it
should be very useful.
- Don
Yamato wrote:
> Don Dailey wrote:
>
>> Go problems don't work for MC programs unless you can arrange them so
>> that finding the right move wins, and anything else loses. I found you
>> can
Yamato wrote:
> Don Dailey wrote:
>
>> Go problems don't work for MC programs unless you can arrange them so
>> that finding the right move wins, and anything else loses. I found you
>> can modify some problems by manipulating the komi to make this true.
>>
>
> I intend to have adjust
Don Dailey wrote:
>Go problems don't work for MC programs unless you can arrange them so
>that finding the right move wins, and anything else loses. I found you
>can modify some problems by manipulating the komi to make this true.
I intend to have adjusted all of them like that (only 1 move =
Go problems don't work for MC programs unless you can arrange them so
that finding the right move wins, and anything else loses. I found you
can modify some problems by manipulating the komi to make this true.
- Don
Yamato wrote:
> Hi all.
>
> I'd like to share my test set for MC programs.
>
37 matches
Mail list logo