Deep NeuroCoevolution

This repository contains the code for the paper A Coevolutionary Approach to Deep Multi-agent Reinforcement Learning.

Paper

Poster

Youtube video

Running the experiments

1. Install the dependencies using either poetry install or pip install -r requirements.txt.
2. Download the required Atari ROMs by running the AutoROM command.
3. If you want to use Weights and Biases for dashboarding, make sure you set the WANDB_API_KEY.
4. Specify the config file you want to use in main.py. You might want to use one of the configs in the configs folder.
5. Run main.py.

If you used Weights and Biases, the metrics and videos should appear in your dashboard: