RLlib Tutorial

These tutorials show you how to use RLlib to train agents in PettingZoo environments.

• PPO for Pistonball: Train a PPO model in a parallel environment

• DQN for Simple Poker Train a DQN model in an AEC environment

RLlib Overview

RLlib is an industry-grade open-source reinforcement learning library. It is a part of Ray, a popular library for distributed ML and scaling python applications.

See the documentation for more information.

• PettingZoo Env

• Implemented Algorithms

Examples using PettingZoo:

Training:

• simple multi-agent: rock-paper-scissors

• multi-agent parameter sharing: waterworld

• multi-agent independent learning: waterworld

• multi-agent leela chess zero

• PR: connect four self-play with pettingzoo

Environments:

• connect four

• chess

Architecture