search

junxiaosong / AlphaZero_Gomoku

An implementation of the AlphaZero algorithm for Gomoku (also called Gobang or Five in a Row)
MIT License
3.27k stars 964 forks source link

Support gomocup protocol #2

Closed tianshuo closed 6 years ago

tianshuo commented 6 years ago

The source code for the protocol can be found here: https://github.com/stranskyjan/pbrain-pyrandom/blob/master/pisqpipe.py

Then it can be used with the Piskvork gomoku manager to compare with other engines like http://www.aiexp.info/pages/yixin.html (which is presently the top gomoku engine)

junxiaosong commented 6 years ago

This repo just serves the purpose of studying the RL principle behind the success of AlphaZero, which learns from pure self-play. In fact, to train a good model for the 15*15 board requires a lot of computation (AlphaZero originally used 5000 TPUs), which is beyond the computational power I have. So there is no plan to make it strong enough so as to compare with top gomoku engines.

tianshuo commented 6 years ago

By playing with other engines, it is possible to estimate the ELO rating of your engine, and see how the ranking of your engine is improving.

On Tue, Dec 26, 2017 at 4:42 PM, junxiaosong notifications@github.com wrote:

This repo just serves the purpose of studying the RL principle behind the success of AlphaZero, which learns from pure self-play. In fact, to train a good model for the 15*15 board requires a lot of computation (AlphaZero originally used 5000 TPUs), which is beyond the computational power I have. So there is no plan to make it strong enough so as to compare with top gomoku engines.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/junxiaosong/AlphaZero_Gomoku/issues/2#issuecomment-353939776, or mute the thread https://github.com/notifications/unsubscribe-auth/AAOL6BObUzOu_uebdzSCD39pxwdYNvQgks5tELGNgaJpZM4RMhmf .

-- 胡天硕 TiansHUo Blog: http://tianshuohu.diandian.com