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. Need a high performance environment for your application? We build them professionally and offer training + extended support. Contact jsuarez馃悺puffer馃悺ai.
All of our documentation is hosted at puffer.ai. @jsuarez5341 on Discord for support. Post there before opening issues. We're always looking for new contributors!