Weaviate Knowledge Cards

Unlock the power of vector search. Our guides will help you conquer vector embeddings and build better AI applications.

Knowledge Cards / Advanced RAG Techniques / Context Compression

Advanced RAG Techniques

Context Compression

Context compression refines RAG retrieval by filtering out irrelevant or redundant data, reducing token usage and costs. It uses a base retriever to find documents, and then a compressor to extract key info. Approaches like embedding-based and lexical-based compression preserve essential content while optimizing efficiency.