概要
This skill provides standardized implementation patterns for managing GPU memory in Python environments, specifically targeting the common issue where Jupyter cells hang after intensive CuPy processing. By forcing the release of CuPy memory pools and triggering explicit Python garbage collection, it ensures that high-performance computing pipelines remain responsive and that VRAM is efficiently returned to the system. It is particularly useful for data scientists and researchers working with large-scale image processing or machine learning workflows where lazy memory cleanup can block execution flow.
