Representation Geometry in Language Models

Ask about this topic

Overview

This field studies the shapes and relationships formed by token embeddings and hidden activations, and how those structures change across layers, contexts, and training. Researchers use linear probes, representational similarity analysis, spectral and manifold methods, sparse autoencoders, and causal interventions such as activation patching and steering. These tools are applied to questions about reasoning, multilingual transfer, retrieval, safety, and reliability, using settings that range from controlled synthetic tasks to benchmarks such as GSM8K and MTEB. A central open tension is that information being decodable or geometrically similar does not guarantee that a model uses it causally; useful structure may also be local, nonlinear, or sensitive to measurement choices.

Latest papers

CardsList

417 papers78 papers in the last four weeks, up 34% on the four weeks before. 0.5% of all new papers.

  1. The Geometry of Emotion Representations Across Scale, Depth, and Instruction Tuning

    Oct 26, 2026Adam HollowellEmotion Representation in Language ModelsRepresentation Geometry in Language Models

    University of North Carolina at Chapel Hill

  2. Five Architectures, One Shared Space Cross-Model Representational Alignment Without Weight Merging

    Oct 24, 2026Oladimeji Anthonio, Dimeji Olawuyi, Oloruntoba Ajayi +1Latent Space AlignmentRepresentation Geometry in Language Models

    Centre for Clinical Intelligence and Safety, `Iy`aw´o · `Iy`aw´o · Faculty of Clinical Sciences, College of Medicine, University of Ibadan

  3. Generalization properties of the token perceptron: the role of embedding and vocabulary

    Oct 23, 2026Dario Bocchi, Matteo Negri, Flavio Nicoletti +1Language Model GeneralizationRepresentation Geometry in Language Models

    Physics Department, University of Rome Sapienza, Rome, Italy · Institute of Nanotechnology, CNR-NANOTEC, Rome, Italy · Laboratoire de Physique Théorique et Modélisation, CY Cergy Paris Université +1

  4. Semantically Rich, Metrically Affect-Blind: A Geometric Blind Spot in Modern Text Embedders

    Oct 23, 2026Mahdi Naser MoghadasiSentence EmbeddingsRepresentation Geometry in Language Models

    BrightMind AI Seattle, WA, USA

  5. A Mathematical Theory of the Jacobian Lens

    Oct 22, 2026Mitchell Ostrow, Daoyuan Qian, Toni Liu +1LLM InterpretabilityRepresentation Geometry in Language Models

    Department of Brain and Cognitive Sciences Massachusetts Institute of Technology · Department of Physics Cornell University

  6. The Geometry of Metaphor: Probing and Steering Metaphor Representations in Large Language Models

    Oct 22, 2026Shihui Li, Mette Andersen, Martha LewisFigurative Language UnderstandingRepresentation Geometry in Language Models

    Institute for Logic, Language, and Computation, University of Amsterdam

  7. Meaning Contracts: Local Semantic Geometry in Large Language Models

    Oct 22, 2026Andrés Daniel Godoy OrtizRepresentation Geometry in Language ModelsLLM Interpretability

    Imperial College London

  8. ECHO: Emergent Common-Mode Hidden Offsets Drive Vocabulary Collapse in Preference Optimization

    Oct 21, 2026Yun Li, Ehsan Javanmardi, Simon Thompson +7Direct Preference OptimizationFine-Tuning Language Models

    Graduate School of Information Science and Technology, The University of Tokyo, Tokyo, Japan · TIER IV, Inc., Tokyo, Japan · Huawei

  9. Language Impairment as a Lens onto Large Language Models

    Oct 20, 2026Muxuan He, Wang Bill Zhu, Yue Chen +5LLM InterpretabilityCircuits in Language Models

    University of Southern California · University of California, Berkeley · Independent Researcher +1

  10. Relative Representations Enable Portable White-Box Monitoring Across Language Models

    Oct 20, 2026Tim Bakker, Christos LouizosLanguage Model ProbingRepresentation Geometry in Language Models

    Qualcomm AI Research

  11. How Language Models Organize Task Information

    Oct 20, 2026Kaiyang Wan, Yuxia WangLanguage Model ProbingLanguage Model Performance Prediction

    INSAIT, Sofia University “St. Kliment Ohridski” Sofia, Bulgaria

  12. When Models Know but Can't Linearly Say: Relational Encoding Limits Linear Vulnerability Classification in Code LLMs

    Oct 19, 2026Rui Melo, Andre Catarino, Claudia Mamede +2LLMs for CybersecuritySoftware Vulnerability Detection

    Carnegie Mellon University, Pittsburgh, PA, USA · Faculty of Engineering, University of Porto, Portugal

  13. Representations in Motion: Layer-wise Rotation and Trajectory Divergence in LLMs

    Oct 18, 2026Zhuonan Yang, Yik Siu Chan, Wannan Yang +1Representation Geometry in Language ModelsDeception in Language Models

    Brown University · New York University

  14. Where Are Language-Specific Features Represented in Multilingual LLMs? Everywhere

    Oct 18, 2026Mikaela Martins, Jonathan DunnMultilingual Language ModelsConcept Localization in Language Models

    Department of Linguistics, University of Illinois Urbana-Champaign

  15. A Symmetry-Matching Approach to Blind-Spot Reduction in Sparse Autoencoders

    Oct 17, 2026Csaba BaloghSparse AutoencodersRepresentation Geometry in Language Models

    National Directorate General for Hospitals, Budapest, Hungary · Data Science in Health Programme, Semmelweis University, Budapest, Hungary

  16. From Stories to Features: Scaling Concept Vectors for AI Alignment & Interpretability

    Oct 17, 2026Dmitry Scherbakov, Oleg Chernikov, Victoria Khamash +5LLM InterpretabilityLLM Alignment

  17. What Shapes Brain Alignment of Large Language Models during Dialogue? Insights from Language Production and Comprehension

    Oct 16, 2026Yuko Nakagi, Yuya Chiba, Yasue KishinoRepresentation Geometry in Language Models

    NTT, Inc.

  18. Knowledge That Is Linear Before Fine-Tuning Generalizes After It

    Oct 15, 2026Faridoun Mehri, Hinrich SchützeLanguage Model GeneralizationFactual Knowledge in Language Models

    Center for Information and Language Processing, LMU Munich Munich Center for Machine Learning (MCML)

  19. Representation Redistribution in Sparse Autoencoders

    Oct 15, 2026Ziye Mei, Haotian MaSparse AutoencodersPrompt Sensitivity

    Department of Computer Sciences, University of Wisconsin–Madison

  20. LLMs program using Reusable and Composable Abstract Representations

    Oct 13, 2026Casper L. Christensen, Kyle J. Ray, Dmitry Manning-Coe +2LLM Code GenerationCompositional Representation Learning

    Simplex · Astera Institute · MATS +1

  21. Embedding Geometry Shapes LLM Routing: Tracing Failure Modes and Improving Generalization

    Oct 13, 2026Leonardo Torti, Emanuele Rimoldi, Tomer Galanti +2LLM RoutingRepresentation Geometry in Language Models

    Massachusetts Institute of Technology. · Texas A&M University.

  22. The Geometry of Language Model Logits: Generic Structure and Learned Departures

    Oct 13, 2026Swathi Shree Narashiman, Aravind Ramana Venkatramanan, Nambi SeshadriAutoregressive Language ModelingRepresentation Geometry in Language Models

    Stanford University · Indian Institute of Technology Madras · University of California San Diego (UCSD)

  23. Probing the Geometry of Color Representations in Multilingual Language Models

    Oct 13, 2026Pablo Loyola, Edison Marrese-Taylor, Andrés Hoyos-IdroboRepresentation Geometry in Language ModelsMultilingual Language Models

    Rakuten Group, Inc. · Graduate School of Engineering, The University of Tokyo, Japan

  24. Everyone Was Right: Lexical Representations in Language Models Are Both Abstract and Holistic

    Oct 13, 2026Zachary Nicholas Houghton, Yu Zhou, Dan Pluth +2Language ModelingRepresentation Geometry in Language Models

    University of Oregon · Vail Systems, Inc.

  25. Gluing Local Contexts into Global Meaning: A Sheaf-Theoretic Decomposition of Transformer Representations

    Oct 12, 2026Bryce Grant, Peng WangTransformer InterpretabilityRepresentation Geometry in Language Models

    Case Western Reserve University

  26. SFT Leaves a Reproducible Signature in Activation Geometry: Calibrating Post-Training Interpretability Against Run-to-Run Noise

    Oct 12, 2026Xupeng Chen, Chenqian Le, Jiayi Cheng +2Language Model Post-TrainingLLM Training

    New York University · University of Southern California · Columbia University

  27. Where Is the "Self" of a Large Language Model? -- From Self-Index to Self-Reference

    Oct 12, 2026Yuchen Li, Xintong Li, Zongqi Fan +2Claim Attribution in Language ModelsLLM Interpretability

    The University of Sydney, Sydney, Australia

  28. Counting with Attention and LayerNorm: Norm Scaling and Representation Geometry

    Oct 12, 2026Ayyub Shaffy, Manoj Dhakal, Safal Shrestha +1Self-AttentionLayer Normalization

    New York University Abu Dhabi

  29. When a Conceptor Complement Becomes Gain: Operator Geometry and Behavioral Attribution in Language Models

    Oct 12, 2026Shuhul Razdan, R BalamuruganLLM InterpretabilityActivation Steering

    School of Computer Science and Engineering (SCOPE), Vellore Institute of Technology, Vellore, India

  30. The Sharp Directions Are Against You: Curvature Analysis of Activation Steering

    Oct 12, 2026Seyedarmin Azizi, Arya Fayyazi, Parsa Razmara +3Language Model SteeringActivation Steering

    University of Southern California, Los Angeles, USA · Intel AI, USA