rag
Retrieval-Augmented Generation, chunking, metadata tags, and retrieval flows.
Файлы1
Открыть в GitHubSKILL.mdskills/ai-ml/rag/SKILL.md
GitHubRag
Retrieval-Augmented Generation, chunking, metadata tags, and retrieval flows.
Core Guidelines
- Chunking Strategy: Size chunks based on document structure (e.g., paragraph or section level).
- Overlap: Include context overlaps to maintain semantic continuity between chunks.
- Re-ranking: Apply re-ranking models on retrieved logs to prioritize top-quality matches.