PufferLib is a fast and sane reinforcement learning library that can train tiny, super-human models in seconds. The included learning algorithm, hyperparameter tuning, and simulation methods are the product of our own research. All our tools are free and open source. Companies can purchase priority service from $10k/month for extended support and tailored advice. We also offer custom simulation engineering, dedicated R&D, and fixed-deliverables for larger projects. Contact jsuarez🐡puffer🐡ai.
The demo below is running live 100% client side in your browser. Hold shift to take control!
Pong
A classic reimagined: Play against our reinforcement learned agent or watch AI vs AI matches. Running at 1M+ steps per second directly in your browser.
Citation
@misc{suarez2024pufferlibmakingreinforcementlearning,
title={PufferLib: Making Reinforcement Learning
Libraries and Environments Play Nice},
author={Joseph Suarez},
year={2024},
eprint={2406.12905},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2406.12905},
}
Contributors
Joseph SuarezFounder & Head Puffer. Writes a lot of code.
David RubinsteinSeveral performance improvements w/ torch compilation, lead pokerl contributor.
Kyoung Whan Choe (최경환)Mujoco bindings, Testing and bug fixes.