StylisticBias is a controlled benchmark for evaluating fine-grained, attribute-level social bias in multimodal large language models (MLLMs). It addresses a key limitation of prior work: most bias studies compare different individuals or demographic groups, making it impossible to disentangle appearance-driven effects from identity-level variation.
StylisticBias keeps identity fixed and varies… See the full description on the dataset page:
https://huggingface.co/datasets/shaghayegh/stylistic-bias-dataset.
