ClovisReranker

ClovisReranker gives you better search results.

In a search pipeline, the first ranking step (vector-based or keyword-based) often returns noisy results. Reranking adds an intelligent second pass that pushes the most relevant results to the top.

🎯 Why reranking?

Vector retrieval is fast but approximate: it returns results that are semantically close, but not always relevant. Reranking refines this initial ranking by evaluating the actual relevance of each document.

The pipeline can be summarized in 3 steps:

1. Search: vector search retrieves a set of documents
2. Re-rank: the reranking model reevaluates each document against the query
3. Answer: only the highest-ranked documents are sent to the LLM

🚀 Ready to improve the relevance of your search results? Check out the full documentation to integrate ClovisReranker into your pipelines:

👉 ClovisReranker documentation