muzero-general
muzero-general copied to clipboard
werner-duvaud
Only One Player: Can we use MuZero?
Search before asking
- [X] I have searched the MuZero issues and found no similar feature requests.
Description
Perfect ideas and codes In my game, there is only one player, constantly moving, assessing the situation, and there is a final reward. How can I use Muzero to optimize AI games? Is there a tutorial for single player training and use?
Additional context
No response
Check out the twenty-one game. It is only one player.
https://github.com/werner-duvaud/muzero-general/blob/master/games/twentyone.py
Hello, thank you to the contributors for their outstanding work on this repository. Regarding the issue you've raised, you might be interested in the project "LightZero: A Unified Benchmark for Monte Carlo Tree Search in General Sequential Decision Scenarios". This repository not only supports the AlphaZero algorithm but also extends support to MuZero and a series of related algorithms and environments, which might meet your requirements. Best wishes.
