迈向透明推理：是什么驱动了大型语言模型的忠实度？
Towards Transparent Reasoning: What Drives Faithfulness in Large Language Models?
AI总结：
本研究探究推理与训练选择如何影响LLM解析的忠实度，在BBQ和MedQA数据集上评估发现few-shot示例、提示策略及指令微调均显著影响模型忠实度，为敏感领域提升可解释性提供见解。
摘要：
大型语言模型（LLMs）经常产生未能忠实反映其预测驱动因素的解析。在医疗保健环境中，这种不忠实性尤其成问题：省略显著临床线索或掩盖虚假捷径的解析会削弱临床医生的信任并导致不安全的决策支持。我们研究推理和训练时的选择如何影响解析的忠实度，重点关注从业者在部署时可以控制的因素。我们在两个数据集——BBQ（社会偏见）和MedQA（医疗执照考试题）上评估了三个LLM（GPT-4.1-mini、LLaMA 70B、LLaMA 8B），并操纵了few-shot示例的数量和类型、提示策略以及训练过程。我们的结果显示：（i）few-shot示例的数量和质量均显著影响模型忠实度；（ii）忠实度对提示设计敏感；（iii）指令微调阶段提高了MedQA上测得的忠实度。这些发现为在敏感领域增强LLM的可解释性和可信度提供了策略见解。
英文摘要：
Large Language Models (LLMs) often produce explanations that do not faithfully reflect the factors driving their predictions. In healthcare settings, such unfaithfulness is especially problematic: explanations that omit salient clinical cues or mask spurious shortcuts can undermine clinician trust and lead to unsafe decision support. We study how inference and training-time choices shape explanation faithfulness, focusing on factors practitioners can control at deployment. We evaluate three LLMs (GPT-4.1-mini, LLaMA 70B, LLaMA 8B) on two datasets-BBQ (social bias) and MedQA (medical licensing questions), and manipulate the number and type of few-shot examples, prompting strategies, and training procedure. Our results show: (i) both the quantity and quality of few-shot examples significantly impact model faithfulness; (ii) faithfulness is sensitive to prompting design; (iii) the instruction-tuning phase improves measured faithfulness on MedQA. These findings offer insights into strategies for enhancing the interpretability and trustworthiness of LLMs in sensitive domains.
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。
