Year: 2026
Type: conference-paper
Abstract: Recent works have analyzed the impact of individual components of neural networks on gendered predictions, often with a focus on mitigating gender bias.However, mechanistic interpretations of gender tend to (i) focus on a very specific gender-related task, such as gendered pronoun prediction, or (ii) fail to distinguish between the production of factually gendered outputs (the correct assumption of gender given a word that carries gender as a semantic property) and gender biased outputs (based on a stereotype).To address these issues, we curate GKnow, a benchmark to assess gender knowledge and gender bias in language models across different types of gender-related predictions.GKnow allows us to identify and analyze circuits and individual neurons responsible for gendered predictions.We test the impact of neuron ablation on benchmarks for disentangling stereotypical and factual gender (DiFair and the test set of GKnow), as well as Stere-oSet.Results show that gender bias and factual gender are severely entangled on the level of both circuits and neurons, entailing that ablation is an unreliable debiasing method.Furthermore, we show that benchmarks for evaluating gender bias can hide the decrease in factual gender knowledge that accompanies neuron ablation.We curate GKnow as a contribution to the continuous development of robust gender bias benchmarks.
Authors: Leonor Veloso, Hinrich Schuetze
Institution: Ludwig-Maximilians-Universität München
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
SDG: Gender equality
Open Access status: hybrid
Funder: Deutsche Forschungsgemeinschaft
Award: G7578079922
OpenAlex ID: W7166852578