ContextCite：将模型生成内容归因于上下文
ContextCite: Attributing Model Generation to Context
AI总结：
针对语言模型生成内容的上下文溯源需求，该研究提出上下文归因问题，并推出可适配任意现有语言模型的简单可扩展方法ContextCite，可用于验证生成陈述、优化回复质量、检测投毒攻击。
摘要：
语言模型在生成回复时如何利用作为上下文提供的信息？我们能否推断出某条生成的陈述究竟是真正基于上下文、属于误解还是凭空捏造的？为解答这些问题，我们提出上下文归因问题：定位导致模型生成特定陈述的上下文片段（若存在的话）。随后我们提出ContextCite，这是一种简单且可扩展的上下文归因方法，可应用于任何现有语言模型之上。最后我们通过三类应用展示ContextCite的实用性：（1）辅助验证生成的陈述；（2）通过剪枝上下文提升回复质量；（3）检测投毒攻击。我们在https://github.com/MadryLab/context-cite 提供了ContextCite的代码。
英文摘要：
How do language models use information provided as context when generating a response? Can we infer whether a particular generated statement is actually grounded in the context, a misinterpretation, or fabricated? To help answer these questions, we introduce the problem of context attribution: pinpointing the parts of the context (if any) that led a model to generate a particular statement. We then present ContextCite, a simple and scalable method for context attribution that can be applied on top of any existing language model. Finally, we showcase the utility of ContextCite through three applications: (1) helping verify generated statements (2) improving response quality by pruning the context and (3) detecting poisoning attacks. We provide code for ContextCite at https://github.com/MadryLab/context-cite.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
