d3rlpy
d3rlpy provides cutting-edge offline and online deep reinforcement learning solutions with an intuitive API suitable for researchers and practitioners. Supporting distributional Q functions and data-parallel distributed training, it is designed for scenarios where online interaction is not feasible. Comprehensive documentation and tutorials assist users, regardless of their deep learning experience. Compatible with Linux, macOS, and Windows, d3rlpy ensures quality through rigorous benchmarking without exaggeration or undue complexity.
