Concept-based interpretability methods offer a lens into the internals of foundation models by decomposing their embeddings into high-level concepts. These concept representations are most useful when they are compositional, meaning that the individual concepts compose to explain the full sample. We show that existing unsupervised concept extraction methods find concepts which are not compositional. To automatically discover compositional concept representations, we identify two salient properties of such representations, and propose Compositional Concept Extraction (CCE) for finding concepts which obey these properties. We evaluate CCE on five different datasets over image and text data. Our evaluation shows that CCE finds more compositional concept representations than baselines and yields better accuracy on four downstream classification tasks. Code and data are available at https://github.com/adaminsky/compositional_concepts .
Funny or Persuasive, but Not Both: Evaluating Fine-Grained Multi-Concept Control in LLMs
University of Cambridge（剑桥大学）；CISPA Helmholtz Center for Information Security（CISPA亥姆霍兹信息安全中心）；Apta（Apta公司）；Royal Holloway, University of London（伦敦大学皇家霍洛威学院）
Cross-Modal Masked Compositional Concept Modeling for Enhancing Visio-Linguistic Compositionality
MoE Key Laboratory of Brain-inspired Intelligent Perception and Cognition, University of Science and Technology of China（中国科学技术大学，教育部脑启发智能感知与认知重点实验室）；Independent Researcher（独立研究员）
Investigating Inner Properties of Multimodal Representation and Semantic Compositionality with Brain-based Componential Semantics
National Laboratory of Pattern Recognition, CASIA（中国科学院自动化研究所模式识别国家重点实验室）；University of Chinese Academy of Sciences（中国科学院大学）；CAS Key Laboratory of Behavioural Science, Institute of Psychology（中国科学院心理研究所行为科学重点实验室）；Department of Psychology, University of Chinese Academy of Sciences（中国科学院大学心理学系）；CAS Center for Excellence in Brain Science and Intelligence Technology（中国科学院脑科学与智能技术卓越创新中心）
