On Mon, Sep 21, 2009 at 08:18:37AM -0700, David Fotland wrote:
> In the original Mogo paper it's the initial value for the children, rather
> than try every child once.
Ah, you mean the First Play Urgency! Thanks, I will try that.
Anyway, I'm happier; after fixing many bugs and improving my playo
2009 1:43 AM
> To: computer-go
> Subject: Re: [computer-go] rave and patterns
>
> On Sun, Sep 20, 2009 at 10:06:01PM -0700, David Fotland wrote:
> > Depends on what you mean by basic UCT. I think I had no UCT priors
> then,
> > just a 1.1 or 1.2 K. The playouts incl
On Sun, Sep 20, 2009 at 10:06:01PM -0700, David Fotland wrote:
> Depends on what you mean by basic UCT. I think I had no UCT priors then,
> just a 1.1 or 1.2 K. The playouts included no self atari, no eye filling,
> no retake ko, and some simple rules for saving group adjacent to last move
> if i
On Tue, Sep 15, 2009 at 08:49:15AM -0700, David Fotland wrote:
> Simple playouts with no eye fills and mogo 3x3 patterns and basic uct beat
> Gnugo 40% (at version 120)
..snip..
> All win rates are on 9x9 vs gnugo 3.7.20 level 10 with 5000 playouts. After
> this I switched to testing 19x19, and st
ses.
From: computer-go-boun...@computer-go.org
[mailto:computer-go-boun...@computer-go.org] On Behalf Of Brian Sheppard
Sent: Thursday, September 17, 2009 11:10 AM
To: computer-go@computer-go.org
Subject: [computer-go] rave and patterns
Olivier and David both: a huge "thank you" f
> David, do I have this right? And is K > N? Or K >> N?
This is somewhat similar in MoGo, with N=5 and K=10 nearly (I say
"nearly" because we have two levels of go expertise, the second being
much more expensive and not yet operationel, so we have in fact K1 and
K2).
> In Pebbles, BTW, the progr
Olivier and David both: a huge "thank you" for sharing your secrets.
I think David makes clear that his large patterns apply only to the UCT
process, and
then only after a significant number of trials are reached. I gather that
the lifecycle
of a node is something like this in MFGO:
1)
Thanks for sharing all this information, David.
> It would be easy to turn off rave and run some tests to do
> the win rate. Would take about a day to get significant
> results. I think RAVE still helps a lot.
I agree that it's easy to turn off rave, but I think that for a fair
comparison
you
I implemented RAVE first.
Simple playouts with no eye fills and mogo 3x3 patterns and basic uct beat
Gnugo 40% (at version 120)
Adding RAVE boosted the win rate to 57% (about 30 more versions of tuning).
I was trying to duplicate the mogo results before adding my own stuff, to
make sure the