AgileRL Tutorial#

These tutorials provide an introductory guide to using AgileRL with PettingZoo. AgileRL’s multi-agent algorithms make use of the PettingZoo parallel API and allow users to train multiple-agents in parallel in both competitive and co-operative environments. This tutorial includes the following:

  • DQN: Train a DQN agent to play Connect Four through curriculum learning and self-play

  • MADDPG: Train an MADDPG agent to play multi-agent atari games

  • MATD3: Train an MATD3 agent to play multi-particle-environment games

AgileRL Overview#

AgileRL is a deep reinforcement learning framework focused on streamlining training for reinforcement learning models. Using evolutionary hyper-parameter optimisation (HPO), AgileRL allows users to train models significantly faster and more accurately when compared with traditional HPO techniques. AgileRL’s multi-agent algorithms orchestrate the training of multiple agents at the same time, and benchmarking has shown up to 4x increase in return in a shorter time-frame when compared with implementations of the very same algorithms in other reinforcement learning libraries.

For more information about AgileRL and what else the library has to offer, check out the documentation and GitHub repo.

Examples using PettingZoo#


Fig1: Performance of trained MADDPG algorithm on 6 random episodes#