揭示大模型内部知识与外部信息的互动机制，发现其常误用外部信息。
Understanding the Interplay between Parametric and Contextual Knowledge for Large Language Models
- 区分内部知识与外部信息的四种关系：支持、互补、冲突、无关。
- 模型在有外部信息时会压制自身知识，即使信息无关或互补。
- 指令优化可部分改善，但难以真正发挥内部知识优势。
大语言模型（LLM）在预训练中存储了大量参数化知识（PK），也可通过引入上下文知识（CK）进行增强。本文探究了PK与CK之间的动态交互，将其关系分为四类：支持性、互补性、冲突性和无关性。为此，我们构建了ECHOQA基准，涵盖科学、事实和常识知识。结果表明，当存在上下文信息时，模型倾向于抑制其内在的参数化知识，即便该信息为互补或无关。尽管定制化指令能促使模型更多依赖内部知识，但仍难以充分发挥其作用。这一发现揭示了大模型在知识密集型任务中的关键脆弱性，引发对其可靠性的担忧。相关资源见https://github.com/sitaocheng/Knowledge_Interplay。
原文摘要 · Abstract (English)
Large language models (LLMs) encode vast amounts of knowledge during pre-training (parametric knowledge, or PK) and can further be enhanced by incorporating contextual knowledge (CK). Can LLMs effectively integrate their internal PK with external CK to solve complex problems? In this paper, we investigate the dynamic interaction between PK and CK, categorizing their relationships into four types: Supportive, Complementary, Conflicting, and Irrelevant. To support this investigation, we introduce ECHOQA, a benchmark spanning scientific, factual, and commonsense knowledge. Our results show that LLMs tend to suppress their PK when contextual information is available, even when it is complementary or irrelevant. While tailored instructions can encourage LLMs to rely more on their PK, they still struggle to fully leverage it. These findings reveal a key vulnerability in LLMs, raising concerns about their reliability in knowledge-intensive tasks. Resources are available at https://github.com/sitaocheng/Knowledge_Interplay
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。