Let's ask a question from HotpotQA — a multi-hop benchmark that requires connecting information across multiple documents. I'll type the question in, select all three retrievers — Dense, BM25, and Hybrid — and hit Run. Now you can instantly see which documents each retriever fetched, their scores, …
Let's ask a question from HotpotQA — a multi-hop benchmark that requires connecting information across multiple documents. I'll type the question in, select all three retrievers — Dense, BM25, and Hybrid — and hit Run. Now you can instantly see which documents each retriever fetched, their scores, and where the overlap is. This alone saves a lot of time when you're debugging a retrieval pipeline.