search

jonathan-laurent / AlphaZero.jl

A generic, simple and fast implementation of Deepmind's AlphaZero algorithm.
https://jonathan-laurent.github.io/AlphaZero.jl/stable/
MIT License
1.23k stars 137 forks source link

What next after training an agent? #187

Closed pepa65 closed 1 year ago

pepa65 commented 1 year ago

I followed the readme in training an agent, except I chose tictactoe (as connect-four ran out of memory), and it completed!

What can I now do to use the result?

jonathan-laurent commented 1 year ago

You can do anything you want with the trained network weights. If you just want to play against the agent or visualize MCTS stats, you can use the explore and play scripts.

pepa65 commented 1 year ago

OK, this works: julia --project -e 'using AlphaZero; Scripts.play("mancala")'