作者位次 #2/4ACL 2026Poster· Natural Language Processing
Diffusion- CAM : Faithful Visual Explanations for d MLLM s
Haomin Zuo · Yidi Li · Luoxiao Yang · Xiaofeng Zhang
摘要
While diffusion Multimodal Large Language Models (dMLLMs) have recently achieved remarkable strides in multimodal generation, the development of interpretability mechanisms has lagged behind their architectural evolution. Unlike traditional autoregressive models that produce sequential activation...