MuZero, first impression how to code it

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
MTaktikos
Posts: 15
Joined: Fri Oct 25, 2019 5:58 pm
Full name: Michael Taktikos

MuZero, first impression how to code it

Post by MTaktikos » Sun Dec 08, 2019 5:56 pm

MuZero is the legacy of AlphaZero and learns to play games (and other decision scenarios) without knowing the rules of the game.
Until now, it was only known that it needs 3 NNs instead of one, as it was in AlphaZero.
Now this site gives new and very impressive information with pseudocode and a lot of Python code, how to build this engine:
https://medium.com/applied-data-science ... 7d5718061a

Post Reply