Dissecting Multimodal In-Context Learning: Modality Asymmetries and Circuit Dynamics in modern Transformers 文章

ArXiv CS.CL2026-05-27NEWSen作者: Yiran Huang, Karsten Roth, Quentin Bouniot, Wenjia Xu, Zeynep Akata
大语言模型
查看原文 →
关系图谱

Dissecting Multimodal In-Context Learning: Modality Asymmetries and Circuit Dynamics in modern Transformers · 相关技术

相关技术

ginORMMultimodal Large Language Models (MLLMs)ODEPLALLMlanguage model递归自我改进多模态statisticsdivide-and-conquer partitioningSSEReferring expression comprehension (REC)Parts-of-Speech (POS) tagsOTANarrative Abstraction Benchmarklarge language modelsIn-context learningHISForFFIEvaluation Driven DevelopmentENAComposed image retrievalCISARGANNtransformersICE