Skip to main content

Knowledge Retrieval Components

Knowledge Retrieval components help your pipeline find the right information for a user query. Use them to turn a question into a search request, fetch the most relevant documents from your database, and prepare the results so downstream components (such as readers or LLMs) have the context they need.

This group includes retrievers for different backends and strategies (for example BM25, embedding, or hybrid retrieval), embedders that create vectors for queries and documents, and utilities to improve or shape retrieval. You can also find components for query expansion, filtering, and ranking or grouping results based on metadata, so you can tune retrieval for quality, speed, and the structure of your data.

Available Components​