Hi Christoph,
I'm currently trying to pretty much replicate this test with my own bot Fluke.
It's more of a confidence test than anything else. I'd like to be sure that
I've got AMAF implemented correctly (if there is any such thing :) ).
myCtest-10k
seems to gain about 420 elo from adding AMAF.
At the moment Fluke (080214 versions) looks to be gaining about 330 elo. In
an effort to be sure I'm comparing like with like, I've just got a couple of
questions about the implementation these versions of myCtest...
*Christoph Birk* /Tue Sep 25 09:57:24 PDT 2007/
On Mon, 24 Sep 2007, Jason House wrote:
>/ I see that myCtest has an AMAF version up on CGOS. Is it possible to share
/>/ details of what was done (under the hood) to get it up and running?
/
'myCtest-xxk' is a pure MC program; no heuristics, no tree, it does
10k (or 50k) light random playouts that do not fill their own 1-point
eyes.
In your "pure MC program", do you use UCB1 to choose the next move to search at
the root? If not, what algorithm are you using? I'm currently using UCB1 for
my test in Fluke.
'myCtest-xxk-AMAF-y' has only one addition. It assigns the random-
playout result not only to the first move, but to a fraction of all
moves (simple all-moves-as-first).
>/ What's different about the -3, -5, and -8 versions? There seems to be
/>/ significant strength differences between them.
/
using N := moves in random simulation
a := AMAF moves
a = pow(N,y)
with
y = 0.25 (3)
y = 0.50 (5)
y = 0.75 (8)
due to the name-length limitation of CGOS (18 characters)
Christoph
A question about the AMAF version... When you are updating the statistics
for the AMAF moves... which moves are you using? e.g. are you using just
the moves of the side to move (e.g. move 1, move 3, move 5.... move n),
or every move (e.g. move 1, move 2, move 3... move n), or some other thing?
When I tried it with every move Fluke seemed to play noticeably weaker than
with every other move.
I also notice that there's a myCtest-10k-AMAF, with no '-y' on the end. Is this
the same as the one with '-8', or is there some other change?
Thanks in advance, Tim.
_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/
