r2l-core

Core traits and data types shared by the r2l reinforcement-learning workspace.

This crate defines environments and spaces, backend-neutral tensors, actors, policies, learning modules, trajectory buffers, and the generic on-policy training loop. It intentionally does not provide a complete end-user training configuration; use r2l for high-level PPO and A2C builders.

API documentation is available on docs.rs.