ACL Findings2023
Debiasing should be Good and Bad: Measuring the Consistency of Debiasing Techniques in Language Models
Robert Morabito, Jad Kabbara, Ali Emami
Abstract
Debiasing methods that seek to mitigate the tendency of Language Models (LMs) to occasionally output toxic or inappropriate text have recently gained traction. In this paper, we propose a standardized protocol which distinguishes methods that yield not only desirable results, but are also consistent with their mechanisms and specifications. For example, we ask, given a debiasing method that is developed to reduce toxicity in LMs, if the definition of toxicity used by the debiasing method is reversed, would the debiasing results also be reversed? We used such considerations to devise three criteria for our new protocol: Specification Polarity, Specification Importance, and Domain Transferability. As a case study, we apply our protocol to a popular debiasing method, Self-Debiasing, and compare it to one we propose, called Instructive Debiasing, and demonstrate that consistency is as important an aspect to debiasing viability as is simply a desirable result. We show that our protocol provides essential insights into the generalizability and interpretability of debiasing methods that may otherwise go overlooked.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
Builds on10
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language ModelsSamuel Gehman, Suchin Gururangan, Maarten Sap, Yejin Choi et al.EMNLP Findings 2020 · 692 citations
- Sustainable Modular Debiasing of Language ModelsAnne Lauscher, Tobias Lüken, Goran GlavasEMNLP Findings 2021 · 85 citations
- Probing Image-Language Transformers for Verb UnderstandingLisa Anne Hendricks, Aida NematzadehACL Findings 2021 · 63 citations
- Beyond Accuracy: Behavioral Testing of NLP Models with CheckListMarco Túlio Ribeiro, Tongshuang Wu, Carlos Guestrin, Sameer SinghACL 2020 · 51 citations
Related papers
- Leashing the Inner Demons: Self-Detoxification for Language ModelsCanwen Xu, Zexue He, Zhankui He, Julian J. McAuleyAAAI 2022 · 30 citations
- Self-Detoxifying Language Models via Toxification ReversalChak Tou Leong, Yi Cheng, Jiashuo Wang, Jian Wang et al.EMNLP 2023 · 12 citations
- Large Language Models can Become Strong Self-DetoxifiersChing-Yun Ko, Pin-Yu Chen, Payel Das, Youssef Mroueh et al.ICLR 2025
- CMD: a framework for Context-aware Model self-DetoxificationZecheng Tang, Keyan Zhou, Juntao Li, Yuyang Ding et al.EMNLP 2024 · 1 citation
- Chinese Toxic Language Mitigation via Sentiment Polarity Consistent RewritesXintong Wang, Yixiao Liu, Jingheng Pan, Liang Ding et al.EMNLP 2025 · 1 citation
