ICLR 2025: Revealing and Reducing Gender Biases in Vision and Language Assistants

The paper “Revealing and Reducing Gender Biases in Vision and Language Assistants (VLAs)” was accepted to ICLR 2025.

This is joint work led by Leander Girrbach with Stephan Alaniz, myself, Trevor Darrell, and Zeynep Akata.

We evaluate gender bias in 22 multimodal LLMs across skills and occupations, and find that fine-tuning-based debiasing methods achieve the best trade-off between debiasing strength and retaining task performance.