Sorry for the slow reply, I had a busy week.
I didn't look much at the code. Both the chess and the go version have a cpp movegen and use tf in the backend, which seems to be the only sane decision at the moment (at least unless you are a big company doing something very unusual).
From the data science perspective they just work along the paper google published.
Who is online
Users browsing this forum: No registered users and 1 guest