Bias A-head? Analyzing Bias in Transformer-Based Language Model Attention Heads

Yi Yang, Hanyu Duan, Ahmed Abbasi, John P. Lalor, KAR YAN TAM

Keywords: pretrained language model, stereotypical biases, BERT, self-attention

TL;DR: A small set of attention heads may contribute to a language model's stereotypical bias.

Abstract:

Transformer-based pretrained large language models (PLM) such as BERT and GPT have achieved remarkable success in NLP tasks.
However, PLMs are prone to encoding stereotypical biases. Although a burgeoning literature has emerged on stereotypical bias mitigation in PLMs, such as work on debiasing gender and racial stereotyping, how such biases manifest and behave internally within PLMs remains largely unknown. Understanding the internal stereotyping mechanisms may allow better assessment of model fairness and guide the development of effective mitigation strategies. In this work, we focus on attention heads, a major component of the Transformer architecture, and propose a bias analysis framework to explore and identify a small set of \textbf{biased heads} that are found to contribute to a PLM's stereotypical bias. We conduct extensive experiments to validate the existence of these biased heads and to better understand how they behave. We investigate gender and racial bias in the English language in two types of Transformer-based PLMs: the encoder-based BERT model and the decoder-based autoregressive GPT model. Overall, the results shed light on understanding the bias behavior in pretrained language models.

Primary Area: societal considerations including fairness, safety, privacy

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 6650