DR-RAG is a two-stage retrieval process which is focussed on calling the LLM only once by performing a multi-hop process on QA datasets.