作者位次 #2/4ACL 2026Findings· Findings of ACL
DOSE : Data Selection for Multi-Modal LLM s via Off-the-Shelf Models
Biao Wu · Yiwu Zhong · Meng Fang · Ling Chen
摘要
High-quality and diverse multimodal data are essential for improving vision–language models (VLMs), yet existing datasets often contain noisy, redundant, and poorly aligned samples. To address these problems, data filtering is commonly used to enhance the efficiency and performance of multimodal...