Hi,
For a monte carlo search, are only valid moves performed? Or does it
work from beginning to the end of a playout using whatever free position
is available on the board?
Because I read here that people can do 25k playouts per second while my
program can only do ~ 20 per second when doing full
On Sat, Mar 28, 2015 at 08:51 +0100, folkert wrote:
> Hi,
>
> For a monte carlo search, are only valid moves performed? Or does it
> work from beginning to the end of a playout using whatever free position
> is available on the board?
I am also interested in the question.
> Because I read here
I am not sure I understand the question. The only thing that is typically
not checked in the playouts is superko. What other "validity checks" are
you performing?
Álvaro.
On Sat, Mar 28, 2015 at 9:54 AM, holger krekel wrote:
> On Sat, Mar 28, 2015 at 08:51 +0100, folkert wrote:
> > Hi,
> >
>
Well you could check if a field is empty or not but you could also
check if you're putting stones in a dead area or if it would be a
suicide, etc.
On Sat, Mar 28, 2015 at 10:15:32AM +0100, ?lvaro Begu? wrote:
> I am not sure I understand the question. The only thing that is typically
> not checked
On Sat, Mar 28, 2015 at 10:15:32AM +0100, Álvaro Begué wrote:
> I am not sure I understand the question. The only thing that is typically
> not checked in the playouts is superko. What other "validity checks" are
> you performing?
There is a few:
(i) No single stone suicide. Can't think of how t
> > Because I read here that people can do 25k playouts per second while my
> > program can only do ~ 20 per second when doing full validity checks on
> > all steps.
> Do you have a reference, some context for the 25K playouts?
Look for the topic "What's a good playout speed?" in the mailinglist
a
hi!
On Sat, Mar 28, 2015 at 11:30:37AM +0100, folkert wrote:
> > > Because I read here that people can do 25k playouts per second while my
> > > program can only do ~ 20 per second when doing full validity checks on
> > > all steps.
> > Do you have a reference, some context for the 25K playouts?
20 playouts is a bit slow Folkert. But I've been there, too! Now my bot is
at around 1000 19x19 pps and 5000 on 9x9 (but I also check to not play into
single point eyes). That's still quite slow and further optimisation will
be required. But OTOH on a 4 core machine without UCT (only the AMAF)
heur
Good to know Petr! Where does the strength "come from"? Sophisticated playouts
or a search algorithm or both?
Von meinem iPhone gesendet
> Am 28.03.2015 um 11:38 schrieb Petr Baudis :
>
> hi!
>
> On Sat, Mar 28, 2015 at 11:30:37AM +0100, folkert wrote:
Because I read here that people can
20 playouts per second is pretty slow. Are you sure youre not just looping
around endlessly for 1 moves or something?
Playing in a dead area is legal, as is filling your own eyes etc. Only suicide,
and ko is illegal. Even superko is a pretty cheap check.
On Sat, Mar 28, 2015 at 11:45:48AM +0100, Urban Hafner wrote:
> Good to know Petr! Where does the strength "come from"? Sophisticated
> playouts or a search algorithm or both?
Frankly, I don't know for sure! The downside of Michi's slowness is
that playtesting takes too many resources so configu
Oh wait, superko check is not that cheap, but it is so rare, you can probably
ignore it in playouts, and jist check befote submitting a move to the server.
If its superko, then jist pass pethaps. ___
Computer-go mailing list
Computer-go@computer-go.org
Exactly you don't check super ko in the playouts and you generally stop
playouts at 3 * boardsize * boardsize. But once you start adding more logic
into the playouts (like not playing in single point eyes) they tend not to hit
this limit.
Von meinem iPhone gesendet
> Am 28.03.2015 um 11:53 sc
One normally checks superko in the UCT tree but not in the playouts.
On Sat, Mar 28, 2015 at 11:53 AM, hughperkins2
wrote:
> Oh wait, superko check is not that cheap, but it is so rare, you can
> probably ignore it in playouts, and jist check befote submitting a move to
> the server. If its su
> But my guess based on ad hoc tests during the development is that the
> contribution of basic playout heuristics and RAVE+priors may be about
> 1:1 (with large pattern priors giving further extra boost).
I'll have to start reading papers on those concepts then, I guess. :)
Thanks Petr!
___
Still, it would be nice if the computer could learn the heuristics
itself, by self-play.
Which is why my bot is still stuck on 25k rating :-D At least, that's
my excuse :-P
On Sat, Mar 28, 2015 at 7:18 PM, Urban Hafner wrote:
>
>> But my guess based on ad hoc tests during the development is th
What does that translate to on CGOS 13x13? I have a hard time estimating how
strong my bot is in "real" terms. How exactly do you measure the strength?
Rated games on KGS?
Von meinem iPhone gesendet
> Am 28.03.2015 um 12:33 schrieb Hugh Perkins :
>
> Still, it would be nice if the computer cou
Well, what I did was connect my bot to kgs a few times, and watch as
it go repeatedly beaten by anything much better than 25k :-)
Normally, there's a few 'randombots' there to start with. Once you can
beat those (which is harder than it sounds, or it is if your program
has bugs, which mine did :-P
> I just realized - you mentioned that you are using Python to run it - if
> you use PyPy instead, you should get about 3x speedup or more.
That was a great suggestion! With PyPy michi now takes less than one minute
per move. Maybe I'll play it some more..
mr. Andrea Carta
__
On Sat, Mar 28, 2015 at 2:34 PM, Hugh Perkins wrote:
> Well, what I did was connect my bot to kgs a few times, and watch as
> it go repeatedly beaten by anything much better than 25k :-)
>
> Normally, there's a few 'randombots' there to start with. Once you can
> beat those (which is harder than
On Sat, Mar 28, 2015 at 6:14 PM, Urban Hafner
wrote:
> On Sat, Mar 28, 2015 at 2:34 PM, Hugh Perkins
> wrote:
>
>> Well, what I did was connect my bot to kgs a few times, and watch as
>> it go repeatedly beaten by anything much better than 25k :-)
>>
>> Normally, there's a few 'randombots' there
You can name name a specific opponent, and then your bot will play against it.
Automatch works, but tends to result in lots of people being forced to play
your bot, and then leaving the game, after the bot took ages to play in some
ridiculous location, which is kind of embarrassing :-P
___
What elements did you like about CGOS and what do you wish for?
I've begun writing a new version from scratch that isn't TCL based.
With the aim for future use and also open source and open to public
commits.
-Josh
___
Computer-go mailing list
Computer-
Can you put in the game description: "really bad, only play if you're
patient" and put in quicker time controls?
s.
On Mar 28, 2015 3:25 PM, "hughperkins2" wrote:
> You can name name a specific opponent, and then your bot will play against
> it.
>
> Automatch works, but tends to result in lots o
Automatch is automatch: the other person doesnt choose to play rhe bot. ___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go
By the way, for mcts you dont need time controls. Each move takes the same
amount of time, since you just do n playouts, and choose n as you like.
I think my playouts took 2s, which was enoufht for 4 playouts i suppose,
but it was in novemebr, dont remember clearly... It was obvious that ad
I can offer you a factor of 2 speedup...
s.
On Mar 28, 2015 7:59 PM, "hughperkins2" wrote:
> By the way, for mcts you dont need time controls. Each move takes the same
> amount of time, since you just do n playouts, and choose n as you like.
>
> I think my playouts took 2s, which was enoufht for
On 1/14/2015 12:14 AM, Urban Hafner wrote:
... I'm currently in the early stages of writing my own Go engine and
right now I'm trying to make my playouts reasonably fast. I've come a
long way in the past few days. Probably not because the payouts are
really fast right now, but because they we
28 matches
Mail list logo
