Unifying Behavioral and Response Diversity for Open-ended Learning in Zero-sum Games

How to run our algorithm?

Create the new environment using:

conda env create -f environment.yml
conda activate diverse_psro

The Real-World Games (https://arxiv.org/abs/2004.09468)

You can now run Real-World Meta-Games (Alpha Star) by executing

python3 spinning_tops_dpp.py

Tools to analyze and visualize results for Real-World Games

python3 plot.py

The Non-Transitive Mixture Model Environment (https://arxiv.org/abs/2103.07927)

python3 non_mixture_model.py

Tools to analyze and visualize results for Real-World Games

python3 analyze_gauss.py

Google Research Football

It is the company's internal code, which cannot be disclosed here due to the copyright issue.

Finally, thanks for the open source implement of https://github.com/diversepsro/diverse_psro !