How Gender Debiasing Affects Internal Model Representations, and Why It MattersDownload PDF

Anonymous

08 Mar 2022 (modified: 05 May 2023)NAACL 2022 Conference Blind SubmissionReaders: Everyone
Paper Link: https://openreview.net/forum?id=LNXORSRJ05
Paper Type: Long paper (up to eight pages of content + unlimited references and appendices)
Abstract: Common studies of gender bias in NLP focus either on extrinsic bias measured by model performance on a downstream task or on intrinsic bias found in models' internal representations. However, the relationship between extrinsic and intrinsic bias is relatively unknown. In this work, we illuminate this relationship by measuring both quantities together: we debias a model during downstream fine-tuning, which reduces extrinsic bias, and measure the effect on intrinsic bias, which is operationalized as bias extractability with information-theoretic probing. Through experiments on two tasks and multiple bias metrics, we show that our intrinsic bias metric is a better indicator of debiasing than (a contextual adaptation of) the standard WEAT metric, and can also expose cases of superficial debiasing. Our framework provides a comprehensive perspective on bias in NLP models, which can be applied to deploy NLP systems in a more informed manner. Our code and model checkpoints are publicly available.
Presentation Mode: This paper will be presented in person in Seattle
Copyright Consent Signature (type Name Or NA If Not Transferrable): Hadas Orgad
Copyright Consent Name And Address: Technion – Israel Institute of Technology, Haifa
0 Replies

Loading