-
To give my contribution (I am a mathematician) I would like to check on the implementation of the MCTS, but I am weak on c++. I ask anyone (possibly without bothering gcp), to:
* direct me to the c…
-
Just opening a place to specifically discuss the chart(s) and such.
@isty2e I took the Vega Lite sample you provided and have been experimenting with it here:
http://zero-test.sjeng.org/view/pla…
-
The similar opening makes up large proportion in the match against 62B , 10 of the first …
-
I'm assuming it's trained off the more recent ones? If so, how many are used?
As an aside: are the scripts used to drive http://zero.sjeng.org/ checked in anywhere? I suspect that would answer all …
-
the application was unable to start correctly 0xc00007b. I meet this problem. I don't know how to play Leela Zero
-
I was watching this match game and it showed W+4.5.
http://zero.sjeng.org/viewmatch/178a8461a877ae2e4e12db74f90d2580da874a1a5e7ae0abf0f94767704c1239
However, I counted by hand and white have 173 te…
-
Hi,
We are a group of EE researchers, trying to implement a new variation of alphago zero.
while looking for a fast implementation of mcts-go, we have encountered in your work.
unfortunately, this …
-
From 311 self-play games I generated so far, I counted 157 draws, 139 white wins, and 15 black wins. This strikes me as odd, considering that the first move advantage should be worth very little for a…
-
I think it would be good idea to put some more statistics to zero.sjeng.org.
For this login for each user would be necessary. Then every user would see how many games he'd uploaded. Maybe people woul…
-
Hi,
I am quite sure the code right now is offered with no guarantees of any sort, anyway my first run after "softlinking" the two directories you hardcoded gave me.
python3 deepmind_mcts.py
Tr…