facebookresearch/BenchMARL

BenchMARL is a library for benchmarking Multi-Agent Reinforcement Learning (MARL). BenchMARL allows to quickly compare different MARL algorithms, tasks, and models while being systematically grounded in its two core tenets: reproducibility and standardization.

76
/ 100
Verified

Built on TorchRL as its execution backend, BenchMARL leverages Hydra for declarative, composable configuration management—enabling systematic hyperparameter sweeps and multi-run experiments across algorithm-task-model combinations. The library integrates with multiple environment ecosystems (VMAS, PettingZoo, MAgent2, SMACv2, MeltingPot) through a unified interface and outputs evaluation data compatible with marl-eval for statistically rigorous comparative analysis.

580 stars and 785 monthly downloads. Available on PyPI.

Maintenance 10 / 25
Adoption 17 / 25
Maturity 25 / 25
Community 24 / 25

How are scores calculated?

Stars

580

Forks

117

Language

Python

License

MIT

Last pushed

Feb 07, 2026

Monthly downloads

785

Commits (30d)

0

Dependencies

6

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/agents/facebookresearch/BenchMARL"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.