This is interesting. I recently built a search tool that needed to locate documents by keyword or by semantics, so I implemented a hybrid search straight away: BM25 + embeddings (from `gte-base`), with a cross-encoder for reranking.
I found that the lexical search was adding nothing; the embeddings alone produced almost identical results for keyword queries. (The re-ranker, however, made a big difference.)
I found that the lexical search was adding nothing; the embeddings alone produced almost identical results for keyword queries. (The re-ranker, however, made a big difference.)
Is this unusual?