作者位次 #6/11ACL 2026Findings· Findings of ACL
P la M : Training-Free Plateau-Guided Model Merging for Better Visual Grounding in MLLM s
Zijing Wang · Yongkang Liu · Mingyang Wang · Ercong Nie · Deyuan Chen · Zhengjie Zhao · Shi Feng · Daling Wang · Xiaocui Yang · Yifei Zhang · Hinrich Schuetze
摘要
Multimodal Large Language Models (MLLMs) rely on strong linguistic reasoning inherited from their base language models. However, multimodal instruction fine-tuning paradoxically degrades this text’s reasoning capability, undermining multimodal performance. To address this issue, we propose a trai...