- Petr's masters paper is good
- i heard on random kgo chat that: uct + RAVE + 3x3 patterns = gets you a 1dan
player
Unfortunately, uct + fast playouts wont get you much further than 25-20 kyo, as
far as i can tell, in my own experience.
(by the way, i was getting 84000 playouts a second on an
Thanks! That's a great data point as it should be easy to run my own bot on
exactly the same hardware
Hmmm, you know what. I bet the playouts i just mentioned were 11x11 playouts...
19x19 should be quite a lot slower...
___
Computer-go mailing
On Mon, Jan 19, 2015 at 12:37 PM, hughperkins2 hughperki...@gmail.com
wrote:
(by the way, i was getting 84000 playouts a second on an amazon
c3.8xlarge, in case you want to compare, which got me around 25-20kyo... )
Thanks! That's a great data point as it should be easy to run my own bot on
Hey everyone,
the playouts of my bot are finally (somewhat) fast enough to start
implementing UCT and I wonder what the best resources are. There's the Senseis
page http://senseis.xmp.net/?UCT as well as the original paper, but I
wonder if there are other papers out there that are newer and maybe
Hey there,
thanks, I'll check out Petr's thesis. That seems like a good overview of
the whole topic. And as far as what to implement. Yes, UCT can only be a
first step, but as I do this in my free time I have to do it is small
chunks. So first UCT, then maybe RAVE, Mogo style patterns, etc.