HoloFair: Unified T2I Fairness Evaluation and Fair-GRPO Debiasing
Provides a unified evaluation and debiasing framework for multidimensional fairness in T2I models, addressing a known bottleneck in bias assessment.
The paper introduces HoloFair, a benchmark for evaluating multidimensional demographic biases in T2I models, and Fair-GRPO, a reinforcement-learning debiasing method. Experiments on SD3.5-Medium show significant fairness improvements while maintaining image quality.
Text-to-Image (T2I) models have made significant strides in visual realism and semantic consistency, yet they often perpetuate and amplify societal biases. Existing evaluation methods typically address only single-dimensional biases, lacking perspectives to uncover model biases at social-related deeper semantic levels. We introduce HoloFair, a comprehensive benchmark framework for multidimensional demographic bias analysis. Built upon our large-scale fairness-oriented dataset and the SpaFreq (Spatial-Frequency) attribute classifier, this framework proposes the Multi-attribute, Group-wise Bias Index (MGBI) metric, designed to assess both intrinsic diversity and conditional biases. Beyond evaluation, we further introduce Fair-GRPO, a reinforcement-learning-based debiasing method that alters the distribution of generative models through a designed multi-objective reward function. E.g., experiments on the SD3.5-Medium model demonstrate that Fair-GRPO significantly improves multidimensional fairness while maintaining high image quality. We also analyze potential reward hacking phenomena and provide corresponding mitigation strategies. Code and dataset are available at https://github.com/1059684669/HoloFair