This release corresponds to the codebase used for the experiments in the arXiv version of the work, presented at the AI4NextG workshop at NeurIPS 2025.
The environment models edge server processing in a simplified manner, abstracting away explicit queueing and service-time effects.