AI RESEARCH
DOSE: Data Selection for Multi-Modal LLMs via Off-the-Shelf Models
arXiv CS.CL
•
ArXi:2604.16979v1 Announce Type: cross 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 learning, but it