1
MAGIC: Multimodal Alignment & Grounding-aware Instruction Coreset for Vision-Language Models
提出MAGIC方法，通过多模态对齐与接地感知的指令核心集，高效提升视觉语言模型训练效果。
arXiv:2605.26004v1 Announce Type: cross Abstract: Instruction tuning of large vision-language models (LVLMs) increasingly depends on massive multimoda…