Alpha Zero is a general reinforcement learning approach that achieved superhuman performance in the games of chess, shogi, and Go through tabula rasa learning from games of self-play, surpassing previous state-of-the-art programs that relied on handcrafted evaluation functions and domain-specific adaptations.
conda create -n rllib-alpha-zero python=3.10
conda activate rllib-alpha-zero
pip install -r requirements.txt
pip install -e '.[development]'