cuml
Opens the skill on ClawHub, which shows its security scan.
Source ClawHub, published by nvidia
Overview
This NVIDIA skill guides an agent working on Python projects built with cuML, NVIDIA's GPU machine learning library, including running existing scikit-learn, UMAP and HDBSCAN code on a graphics card without changes. Because cuML drops old APIs often, it tells the agent to check the installed version rather than rely on memory. It covers checking that work actually used the GPU, memory recipes for large UMAP and HDBSCAN jobs, and exporting trained models.
Use cases
- Machine learning research
- GPU machine learning
- scikit-learn acceleration
- UMAP
- HDBSCAN
