作者位次 #4/9ACL 2026Poster· Natural Language Processing
Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLM s
Tingchao Fu · Wenkai Wang · Fanxiao Li · Huadong Zhang · Jinhong Zhang · Dayang Li · Yunyun Dong · Renyang Liu · Wei Zhou
摘要
Although Knowledge Editing provides an efficient mechanism for updating the knowledge of Multimodal Large Language Models (MLLMs), we find that current paradigms still suffer from an important yet remain underexplored issue : editing decoupling failure, where entity-related knowledge can be updat...