Some Modalities are More Equal Than Others: Decoding and Architecting Multimodal Integration in MLLMs
某些模态比其他模态更平等:在MLLMs中解码和架构多模态整合
机构 * Boston University(波士顿大学) ; Google DeepMind(谷歌DeepMind)
专题命中 视觉定位与Grounding :grounding(abstract);multimodal large language model(abstract);分类 cs.CV
AI总结 本文研究了多模态大语言模型对矛盾模态的鲁棒性,提出模态对齐调优策略以提升多模态推理能力。