Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model

Paper: https://arxiv.org/abs/1911.08265

This directory contains an implementation of the Pseudocode description of the MuZero algorithm (https://arxiv.org/src/1911.08265v1/anc/pseudocode.py).

The implementation uses SEED RL for scalable RL training.

Pull Requests

At this time, we do not accept pull requests. We are happy to link to forks that add interesting functionality.

Prerequisites

We require tensorflow and other supporting libraries. Tensorflow should be installed separately following the docs.

SEED RL should be installed following instructions here.

To install the other dependencies use

pip install -r requirements.txt

Training

Follow instructions from the SEED repo to run Local Machine Training or Distributed Training.

This directory adds a tictactoe environment and an atari environment. These can be used as the $ENVIRONMENTS when running the seed_rl scripts.

This directory also adds a muzero agent which can be used as the $AGENTS when running the seed_rl scripts.