Once-For-All: A Train-Once and Select-Anytime Framework for Multimodal Instruction Tuning 文章

ArXiv CS.CV2026-05-27NEWSen作者: Mingkang Dong, Hongyi Cai, Xiwen Lei, Jie Li, Tao Zhang, Muxin Pu
大语言模型
查看原文 →
关系图谱

Once-For-All: A Train-Once and Select-Anytime Framework for Multimodal Instruction Tuning · 相关技术

相关技术

ORMODElanguage model多模态target modelZenVision Language ModelsUCTStraight-Through EstimatorStanSpatial Pivot-Aligned Coordinate-free Embedding (SPACE)SPAReferring expression comprehension (REC)Parts-of-Speech (POS) tagsVLMHISForFFIDAPTCLIARGANN