Overview

In machine learning, linear probing means freezing a model’s representations and training a simple classifier or regressor—often logistic regression or ridge regression—to measure which task-relevant features they make accessible. Researchers apply it across layers, modules, and subspaces in vision, language, audio, medicine, and scientific models, using benchmarks for classification, geometry, factual knowledge, and out-of-distribution transfer. Methods now include layer fusion, low-rank and whitened probes, and probe-based steering, while comparisons with fine-tuning help assess label efficiency. A central tension is that easy linear decoding does not by itself show that a model uses that information causally, and probe results can be sensitive to label noise, task design, and distribution shift.

Latest papers

CardsList

42 papers5 papers in the last four weeks, up 25% on the four weeks before. 0.0% of all new papers.

  1. Pretty Linear Liars: Unregularized Probes Favor Intervention-Unfaithful Features

    Nov 16, 2026Karl Vilhelmsson Emneby, Julian Yocum, Cameron AllenLinear ProbingMechanistic Interpretability

  2. Do Probing and Steering Favor the Same Directions Within and Across Language Models?

    Nov 14, 2026Afshin Sadeghi, Daniel BeckerLanguage Model SteeringActivation Steering

  3. Probing Transformer Representations Layer by Layer, Module by Module

    Nov 8, 2026Ambroise Odonnat, Theo Gnassounou, Laetitia Chapel +2Transformer InterpretabilityVision Foundation Models

  4. 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

  5. Is a Linear Probe Evidence of a Linear Representation?

    Oct 15, 2026Nadir Kutluozen, Christian Internò, David KlindtRepresentation Geometry in VLMsLinear Representation Hypothesis

  6. Whitening Improves Robustness to Spurious Correlations in Linear Probes

    Sep 30, 2026Floris Holstege, Bram Wouters, Noud van Giersbergen +1Linear ProbingFeature Whitening

  7. Capacity, Responsiveness and Alignment: What Makes a Latent Structure Actionable

    Sep 28, 2026Or Shafran, Mor GevaCausal Interventions in Language ModelsCausal Interpretability

  8. CG-Probes: Recovering Guardrail Directions from Patient Query Embeddings

    Sep 25, 2026Marko Řeháček, Vítězslav Dušek, Martin Rusinko +1LLM GuardrailsLinear Probing

  9. Exact semantic readout from compressed vector representations

    Sep 16, 2026Daniel QuigleyLinear ProbingEmbedding Compression

  10. The Hallucination Signal Is a Mean Shift: Why Simple Probes Suffice

    Aug 28, 2026Jungseob Lee, Jaehyung Seo, Heuiseok LimLLM Hallucination DetectionHidden-State Probing

  11. How Do Linear Probes Emerge? A Circuit-Tracing Framework with Concept-Targeted Attribution

    Aug 27, 2026Vedant Palit, Florent Draye, Terry Jingchen Zhang +2Mechanistic InterpretabilityFeature Attribution

  12. Measuring Concept Content in Text from LLM Activations: ESG Evidence from Concept Vectors and Linear Probes

    Aug 7, 2026Luc Hazenoot, Zhaochun Ren, Amirhossein ZohrehvandLinear ProbingLLM Interpretability

  13. Training-Free versus Training-Based Intent Classification in LLMs: Accuracy, Robustness, and Failure Modes

    Aug 3, 2026Nan Chen, Zhouhao Yang, Soufiane HayouZero-Shot LearningLLM Evaluation

  14. Foundation Models for Face Presentation Attack Detection: A Unified Linear-Probing Benchmark

    Jul 29, 2026Peter Lorenz, Anjith George, Sébastien MarcelVision Foundation ModelsLinear Probing

  15. Understanding Geometric Representations in Self-Supervised Vision Transformers via Subspace Intervention

    Jul 2, 2026Weichen Zhou, Yawen Zou, Chunzhi Gu +3Neural Representation GeometryLinear Probing

  16. Primary ICD Category Prediction using LLM-based Probing

    Jun 27, 2026Chengyuan Liu, Xinyue Zhang, Yao Li +1Automated Medical CodingLinear Probing

  17. Comparing Linear Probes with Mahalanobis Cosine Similarity

    Jun 17, 2026Zhuofan Josh Ying, Peter Hase, Nikolaus KriegeskorteLinear ProbingMahalanobis Distance

  18. When LLMs Learn to Be Consistently Wrong: A Multi-Model Study of Linear Representations of Synthetic Deception

    May 28, 2026Vahideh ZolfaghariLinear ProbingLanguage Model Safety Evaluation

  19. Optimal Representation Size: High-Dimensional Analysis of Pretraining and Linear Probing

    May 19, 2026Valentina Njaradi, Clémentine Dominé, Rachel Swanson +2Neural Network GeneralizationLinear Probing

  20. Probing for Representation Manifolds in Superposition

    May 18, 2026Alexander ModellFeature SuperpositionNeural Representation Geometry

  21. Polar probe linearly decodes semantic structures from LLMs

    May 13, 2026Pablo J. Diego-Simón, Pierre Orhan, Emmanuel Chemla +2Representation LearningLinear Probing

  22. Closed-Form Linear-Probe Dataset Distillation for Pre-trained Vision Models

    May 8, 2026Bincheng Peng, Guang Li, Ping Liu +2Linear ProbingDataset Distillation

  23. Tracing Relational Knowledge Recall in Large Language Models

    Apr 21, 2026Nicholas Popovič, Michael FärberLinear ProbingFactual Knowledge in Language Models