How faithful are RAG models? Quantifying the tug-of-war between RAG and LLMs' internal prior

arXiv:2404.10198v1 Announce Type: new
Abstract: Retrieval augmented generation (RAG) is often used to fix hallucinations and provide up-to-date knowledge for large language models (LLMs). However, in cases when the LLM alone incorrectly answers a question, does providing the correct retrieved content always fix the error? Conversely, in cases where the retrieved content is incorrect, does the LLM know to ignore the wrong information, or does it recapitulate the error? To answer these questions, we systematically analyze the tug-of-war between a …

abstract arxiv cases cs.ai cs.cl error hallucinations however knowledge language language models large language large language models llm llms prior question rag retrieval retrieval augmented generation type war