arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2501.06254cs.CLcs.AIcs.LG
收藏

通过多义词的表征重新思考稀疏自编码器的评估

Rethinking Evaluation of Sparse Autoencoders through the Representation of Polysemous Words

The University of Tokyo(东京大学)

Gouki Minegishi, Hiroki Furuta, Yusuke Iwasawa, Yutaka Matsuo

AI总结:

针对传统指标忽视 SAE 语义表征能力的问题,本文提出以多义词为核心的单义特征评估套件,揭示 MSE-L0 Pareto 优化的局限及 LLM 层与注意力机制的消歧作用。

摘要:

稀疏自编码器(SAEs)作为一种很有前景的工具受到了广泛关注,它通过将多义神经元的复杂叠加映射为单义特征并构建稀疏的词语词典,来提升大型语言模型(LLMs)的可解释性。然而,均方误差和 L0 稀疏度等传统性能指标忽略了对 SAEs 语义表征能力的评估,即它们能否在保留词语语义关系的同时获得可解释的单义特征。例如,一个学习到的稀疏特征能否区分同一个词的不同含义,目前并不明确。本文中,我们提出了一套针对 SAEs 的评估方法,通过聚焦多义词来分析单义特征的质量。研究发现,为改善 MSE-L0 Pareto 前沿而开发的 SAEs 可能会混淆可解释性,这并不必然增强单义特征的提取。利用多义词分析 SAEs 还可以揭示 LLMs 的内部机制;更深层和 Attention 模块有助于区分一个词的多义性。我们以语义为中心的评估为多义性和现有 SAE 目标提供了新的见解,并有助于开发更实用的 SAEs。

英文摘要:

Sparse autoencoders (SAEs) have gained a lot of attention as a promising tool to improve the interpretability of large language models (LLMs) by mapping the complex superposition of polysemantic neurons into monosemantic features and composing a sparse dictionary of words. However, traditional performance metrics like Mean Squared Error and L0 sparsity ignore the evaluation of the semantic representational power of SAEs -- whether they can acquire interpretable monosemantic features while preserving the semantic relationship of words. For instance, it is not obvious whether a learned sparse feature could distinguish different meanings in one word. In this paper, we propose a suite of evaluations for SAEs to analyze the quality of monosemantic features by focusing on polysemous words. Our findings reveal that SAEs developed to improve the MSE-L0 Pareto frontier may confuse interpretability, which does not necessarily enhance the extraction of monosemantic features. The analysis of SAEs with polysemous words can also figure out the internal mechanism of LLMs; deeper layers and the Attention module contribute to distinguishing polysemy in a word. Our semantics focused evaluation offers new insights into the polysemy and the existing SAE objective and contributes to the development of more practical SAEs.

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。

↑