KeepGPU is a small, polite GPU keeper for shared machines. It allocates only when utilization backoff permits, keeps the requested VRAM signal lightweight, and releases cleanly when you are done.
pip install keep-gpu
keep-gpu --gpu-ids 0 --vram 1GiB --interval 60Press Ctrl+C to release. CUDA, ROCm/HIP, and Apple Silicon MPS are supported
when PyTorch can see the target device.
- Documentation is the full guide hub.
- Getting Started covers install options and the first hardware check.
- Citation has the BibTeX entry.