[Feature Request] Muzero and MCTS implementations

pytorch / rl

A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.

https://pytorch.org/rl

MIT License

2.25k stars 297 forks source link

[Feature Request] Muzero and MCTS implementations #1845

Open Prakyathkantharaju opened 8 months ago

Prakyathkantharaju commented 8 months ago

Motivation

It would be great to have an MCTS and Alphazero implementation, including other model-based RL for benchmarking and comparison.

Solution

I can write a loss function of this policy.

Alternatives

There are limited RL libraries that have a base implementation of Muzero.

Additional context

None.

Checklist

[x] I have checked that there is no similar issue in the repo (required)

vmoens commented 8 months ago

Interestingly someone just dropped a suggestion to help us implement alpha zero https://github.com/pytorch/rl/discussions/1844 If you want to collaborate or follow the progress, i'd suggest to join our discord challenge here, I just created an MCTS channel!