You have an environment, a PyTorch model, and a reinforcement learning library that are designed to work together but don’t. PufferLib provides one-line wrappers that make them play nice.

All of our Documentation is hosted by github.io. @jsuarez5341 on Discord for support -- post here before opening issues. I am also looking for contributors interested in adding bindings for other environments and RL frameworks.