Preprint
Oct 2026
CoTA++: Understanding the Repeat Curse in dMLLMs from an Information Flow Perspective
Diffusion-based multimodal large language models (dMLLMs) decode by iterative denoising with bidirectional attention, which precludes the standard autoregressive key--value cache. Practical deployments therefore rely on approximate caching that reuses intermediate states across denoising steps. Such caching, however, o...
Qi-Yan Zhao, Xiao-Feng Zhang, Xu-Yao Zhang et al.
· 0 citations