hybrid search in hyde

[Mycroft-s]

thanks for your contribution. we want to use [dense, sparse] hybrid search in our RAG system. And, there is a question, does the contriever only use to embed the generated doc to vector and run the retrieval part to find the most related real doc? could i understand that hyde replace the query rewriting part? query rewriting is for a more detailed and integrated query, but hyde use query and LLM to find related documents driectly. do we can use our embedding model and retrieval cmponent to use for the contriever?