Preprint
Aug 2026
VADER: Adaptive Debiasing for Hallucination Mitigation in Video Large Language Models
Video-Adaptive Debiasing via Evidence Reweighting (VADER), a training-free framework with two complementary modules that combines Contrastive decoding then down-weights predictions that remain confident after selective evidence erasure.
Dong Xing, Jiaxin Chen, Hang Yang et al.
· 0 citations