UDE is an open-source Python library for virtualizing reinforcement learning environments. It supports seamless offload and provides a uniform API to control the flow and life-cycle of the simulation environments.
To install the UDE toolkit, run the pip install udecommand.
Python 3.6, 3.7, 3.8, and 3.9 are supported on Linux, Windows, and macOS.
UDE whitepaper is available at https://arxiv.org/abs/2205.06946. If you reference or use UDE in your research, please cite:
@misc{la2022unified,
title={Unified Distributed Environment},
author={Woong Gyu La and Sunil Muralidhara and Lingjie Kong and Pratik Nichat},
year={2022},
eprint={2205.06946},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
The source code is released under Apache 2.0.