标签
This paper investigates how epistemic stance (qualifiers, attributions) survives memory compression in AI agent memory systems. It finds that making the stance explicit as a labelled field improves retention significantly, while merely lengthening the text does not.
本文介绍了认知立场灵活性探测(ESFP),这是一个行为基准,用于测量大语言模型在将主张归因于专家与表达自身立场之间如何转换其认知语域。通过评估八款前沿模型,作者发现认知灵活性在很大程度上与通用能力正交,其中立场内容密度提供了最强的信号。