contextual retrieval vs standard RAG

Topics

• RAG

Anthropic propses using contextualized chunking in RAG systems where LLMs create context for the chunks. This method is simple (but adds latency and cost), but gives better results.

Related

• contextualized chunking