arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2405.00301cs.CL
收藏

基于可学习干预与不确定性表达的大语言模型真实性增强方法

Enhanced Language Model Truthfulness with Learnable Intervention and Uncertainty Expression

University of Michigan(密歇根大学)

Farima Fatahi Bayat, Xin Liu, H. V. Jagadish, Lu Wang

AI总结:

本文提出可学习干预方法LITO,通过自动调整每个查询上下文的最优干预强度并引入不确定性表达下的拒绝回答机制,在保持任务准确率的同时提升大语言模型的真实性。

摘要:

大语言模型(LLMs)能够生成长篇且连贯的文本,但经常虚构事实,这削弱了它们的可靠性。为缓解这一问题,推理时方法将LLM的表征引导至此前为真实性激发而学习的“真实性方向”。然而,以相同强度应用这些真实性方向无法在不同查询上下文中泛化。我们提出LITO,一种用于真实性优化的可学习干预方法,能够自动识别针对每个具体上下文定制的最优干预强度。LITO基于逐步增加的干预强度水平探索一系列模型生成结果。它选择最准确的回答,或在预测高度不确定时拒绝回答。在多个LLM和问答数据集上的实验表明,LITO在保持任务准确率的同时提升了真实性。LITO的自适应特性克服了“一刀切”干预方法的局限性,仅在模型自信时通过反映其内部知识来最大化真实性。我们的代码可在https://github.com/launchnlp/LITO获取。

英文摘要:

Large language models (LLMs) can generate long-form and coherent text, yet they often hallucinate facts, which undermines their reliability. To mitigate this issue, inference-time methods steer LLM representations toward the "truthful directions" previously learned for truth elicitation. However, applying these truthful directions with the same intensity fails to generalize across different query contexts. We propose LITO, a Learnable Intervention method for Truthfulness Optimization that automatically identifies the optimal intervention intensity tailored to each specific context. LITO explores a sequence of model generations based on increasing levels of intervention intensities. It selects the most accurate response or refuses to answer when the predictions are highly uncertain. Experiments on multiple LLMs and question-answering datasets demonstrate that LITO improves truthfulness while preserving task accuracy. The adaptive nature of LITO counters the limitations of one-size-fits-all intervention methods, maximizing truthfulness by reflecting the model's internal knowledge only when it is confident. Our code is available at https://github.com/launchnlp/LITO.

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。

↑