3 papers
cs.MM2026
Mechanism-Level Evaluation for Vision-Language Models: Controlled Activation-Replacement Diagnosis of Gender Bias
Zhipeng Zhao, Wenxu Wang, Peishun Liu +1
Behavioral benchmarking reveals \emph{what} biases exist in vision-language models but not \emph{which internal components} are most sensitive to targeted intervention, precluding…
cs.CV2026
ViD: Vision-Dominant Gender Bias Mitigation for Large Vision-Language Models
Zhipeng Zhao, Zhaoqiang Wei, Peishun Liu +2
Gender bias in large vision-language models (LVLMs) undermines their fairness and reliability, compromising output trustworthiness. Current mitigation methods rely on training-phas…
cs.CV2026
What Do Hallucinations Reveal About Multimodal Reasoning? Diagnosing Visual Grounding Failures via Contrastive Decoding Probes
Zhipeng Zhao, Wenxu Wang, Peishun Liu +1
When strong multimodal models are widely available, progress requires new scientific methodologies beyond benchmark scores---using models as instruments for understanding behavior.…