arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.07578cs.CL
收藏

句法扰动揭示预训练语言模型中层次短语结构的表征相关性

Syntactic Perturbations Reveal Representational Correlates of Hierarchical Phrase Structure in Pretrained Language Models

Columbia University(哥伦比亚大学);Intel Labs(英特尔实验室);MIT-IBM Watson AI(MIT-IBM沃森人工智能实验室);Massachusetts Institute of Technology(麻省理工学院)

Matteo Alleman, Jonathan Mamou, Miguel A Del Rio, Hanlin Tang, Yoon Kim, SueYeon Chung

摘要:

虽然来自预训练语言模型的基于向量的语言表示为许多NLP任务设定了新标准,但尚未对其内部工作原理进行完整的解释。特别是,我们还不完全清楚这些表示捕捉了哪些方面的语法,也不清楚它们是如何(如果有的话)沿着网络的堆叠层构建的。在本文中,我们的目标是解决这样的问题与一般类的干预,输入扰动为基础的分析表示从预训练的语言模型。从计算和认知神经科学的表征不变性的概念,我们进行了一系列的探测器设计,以测试这些表征的几种结构的句子的敏感性。每一个探测都涉及交换句子中的单词,并将扰动句子的表征与原始句子进行比较。我们用三种不同的扰动进行实验:(1)随机排列不同宽度的n-gram,以测试表征对单词位置敏感的程度;(2)交换两个跨度,它们形成或不形成句法短语,以测试对全局短语结构的敏感性;以及(3)交换两个相邻的词,它们分开或不分开句法短语,以测试对局部短语结构的敏感性。这些探测的结果共同表明,Transformers沿着它们的层建立对句子较大部分的敏感性,并且层次短语结构在这个过程中起着作用。更广泛地说,我们的研究结果还表明,结构化输入扰动扩大了可以在通常不透明的深度学习系统上执行的分析范围,并且可以作为现有工具(如监督线性探针)的补充,用于解释复杂的黑盒模型。

英文摘要:

While vector-based language representations from pretrained language models have set a new standard for many NLP tasks, there is not yet a complete accounting of their inner workings. In particular, it is not entirely clear what aspects of sentence-level syntax are captured by these representations, nor how (if at all) they are built along the stacked layers of the network. In this paper, we aim to address such questions with a general class of interventional, input perturbation-based analyses of representations from pretrained language models. Importing from computational and cognitive neuroscience the notion of representational invariance, we perform a series of probes designed to test the sensitivity of these representations to several kinds of structure in sentences. Each probe involves swapping words in a sentence and comparing the representations from perturbed sentences against the original. We experiment with three different perturbations: (1) random permutations of n-grams of varying width, to test the scale at which a representation is sensitive to word position; (2) swapping of two spans which do or do not form a syntactic phrase, to test sensitivity to global phrase structure; and (3) swapping of two adjacent words which do or do not break apart a syntactic phrase, to test sensitivity to local phrase structure. Results from these probes collectively suggest that Transformers build sensitivity to larger parts of the sentence along their layers, and that hierarchical phrase structure plays a role in this process. More broadly, our results also indicate that structured input perturbations widens the scope of analyses that can be performed on often-opaque deep learning systems, and can serve as a complement to existing tools (such as supervised linear probes) for interpreting complex black-box models.

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。

↑