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 2311.08968cs.CLcs.AI
收藏

识别大型语言模型中的线性关系概念

Identifying Linear Relational Concepts in Large Language Models

University College London(伦敦大学学院)

David Chanin, Anthony Hunter, Oana-Maria Camburu

AI总结:

本文提出线性关系概念(LRC)技术,通过线性关系嵌入(LRE)及其反转在Transformer语言模型潜在空间中定位人类可解释概念的方向,在概念分类和因果改变模型输出方面优于黑盒探测分类器。

摘要:

Transformer语言模型(LM)已被证明可以将概念表示为隐藏激活潜在空间中的方向。然而,对于任何人类可解释的概念,我们如何找到它在潜在空间中的方向?我们提出了一种称为线性关系概念(LRC)的技术,通过首先将主体与客体之间的关系建模为线性关系嵌入(LRE),来寻找与人类可解释概念相对应的概念方向。我们发现,反转LRE并使用较早的客体层,可以得到一种强大的概念方向寻找技术,其性能优于标准的黑盒探测分类器。我们评估了LRC作为概念分类器的性能,以及它们因果改变模型输出的能力。

英文摘要:

Transformer language models (LMs) have been shown to represent concepts as directions in the latent space of hidden activations. However, for any human-interpretable concept, how can we find its direction in the latent space? We present a technique called linear relational concepts (LRC) for finding concept directions corresponding to human-interpretable concepts by first modeling the relation between subject and object as a linear relational embedding (LRE). We find that inverting the LRE and using earlier object layers results in a powerful technique for finding concept directions that outperforms standard black-box probing classifiers. We evaluate LRCs on their performance as concept classifiers as well as their ability to causally change model output.

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。

↑