Skill UI

NeMo Curator runs GPU-accelerated data curation for LLM training, covering text, image, video, and audio pipelines with fuzzy/semantic deduplication, quality classifiers, and PII/NSFW filters before scaling across RAPIDS clusters.

Extract structured, quality-scored domain knowledge from live AI sessions or local open-source models via Ollama, producing searchable reference datasets and traditional ML-ready outputs without needing extra API keys.

Produces publication-quality figures for ML/AI conference papers by extracting system components for Gemini-based architecture diagrams and turning experimental data into matplotlib/seaborn charts with auto-selected types, layouts, and highlights so visuals match the paper context.