Language Bias in LVLMs: From In-Depth Analysis to Simple and Effective Mitigation 文章

ArXiv CS.CL2026-05-26NEWSen作者: Yangneng Chen, Jing Li
大语言模型
查看原文 →
关系图谱

Language Bias in LVLMs: From In-Depth Analysis to Simple and Effective Mitigation · 相关技术

相关技术

ODELLMlanguage model多模态alignmentUCTStraight-Through EstimatorStanReferring expression comprehension (REC)OWSNPUNATMITLarge vision-language modelVision-Language ModelsVLMLVLMslarge language modelsHISEffort Metric AttentionDirect Preference OptimizationDPOARGANNTPU