Berget AI Reranker icon

Berget AI Reranker

Rerank documents retrieved from a Vector Store using a Berget AI reranker model. Plugs into Vector Store retrievers and other nodes that accept an AiReranker connection.

Berget AI Reranker

AI-generated

Summary

Rerank a larger set of documents retrieved by a parent Vector Store retriever node using a selected Berget AI reranker model, returning the top N most relevant documents based on the query.

Inputs

  • model (required) — Select the Berget AI reranker model to use, loaded live from the Berget API.
  • topN — Set the maximum number of documents to keep after reranking; controls how many documents survive the rerank step from the larger candidate set passed by the parent node.
  • options.timeout — Specify the maximum time in milliseconds to wait for the rerank API response; defaults to 60000 ms.

Output shape

a NodeConnectionTypes.AiReranker output connection producing a reranker instance to be consumed by a connected Vector Store retriever or compatible node

This node cannot be run standalone and must be connected as a reranker to a parent node accepting AiReranker input. The output is a reranker instance encapsulating the selected model, which the parent node uses internally to compress (rerank) its retrieved documents before final delivery.

Links

Discussion