Large study finds RAG may not improve biomedical QA

Arxiv CS.CL (NLP/language models)

Study of retrieval-augmented generation for medical question answering shows retrieval does not boost accuracy and can even hurt. Contradicts prior claims of substantial gains.

rag

Arxiv CS.CL (NLP/language models)·Erfan NourbakhshRocky SlavinKe YangAnthony Rios

7 days ago