探究大语言模型忠实自解释的训练与泛化
Investigating Training and Generalization in Faithful Self-Explanations of Large Language Models
AI总结：
本研究通过训练提升大语言模型的自解释忠实性，并验证其在不同任务和风格中的泛化能力。
摘要：
大语言模型有能力根据用户指令以多种风格生成对自己预测的解释。最近的研究探讨了这些自解释是否忠实反映模型的实际行为，并发现它们通常缺乏忠实性。然而，如何提高忠实性的问题仍被忽视。此外，由于不同解释风格在表面特征上有所不同，不清楚在一种风格中观察到的改进是否在其他风格中也会出现。本研究分析了训练忠实自解释的效果以及这些效果的泛化程度，使用三个分类任务和三种解释风格。我们使用特征归因方法构建了一字受限解释，这些解释很可能忠实，然后用这些伪忠实自解释对指令调优模型进行持续学习。我们的实验表明，训练可以提高所有分类任务和解释风格的自解释忠实性，并且这些改进在多字设置和未见过的任务中显示出泛化的迹象。此外，我们发现三种风格之间有持续的跨风格泛化，表明训练可能有助于更广泛地提高忠实自解释能力。
英文摘要：
Large language models have the potential to generate explanations for their own predictions in a variety of styles based on user instructions. Recent research has examined whether these self-explanations faithfully reflect the models' actual behavior and has found that they often lack faithfulness. However, the question of how to improve faithfulness remains underexplored. Moreover, because different explanation styles have superficially distinct characteristics, it is unclear whether improvements observed in one style also arise when using other styles. This study analyzes the effects of training for faithful self-explanations and the extent to which these effects generalize, using three classification tasks and three explanation styles. We construct one-word constrained explanations that are likely to be faithful using a feature attribution method, and use these pseudo-faithful self-explanations for continual learning on instruction-tuned models. Our experiments demonstrate that training can improve self-explanation faithfulness across all classification tasks and explanation styles, and that these improvements also show signs of generalization to the multi-word settings and to unseen tasks. Furthermore, we find consistent cross-style generalization among three styles, suggesting that training may contribute to a broader improvement in faithful self-explanation ability.
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。
