mlsanityData to 7 Oct 2026

Probing for Representation Manifolds in Superposition

arXiv 2605.18537 ↗First posted 18 May 2026cs.LG cs.AI stat.MLcited by 0

Not on the map. Why?

Abstract (arXiv)

This paper introduces the Manifold Probe, a supervised method for discovering representation manifolds in superposition. The method generalizes linear regression probes by learning the space of features of a concept that can be linearly predicted from the representations, and then learning the directions used to encode them. We demonstrate the probe on representations of time and space in Llama 2-7b, finding manifolds which linearly represent an interpretable set of features in each case. In the case of time, we show that by steering along the manifold, we can influence the model's completions about the years in which famous songs, movies and books were released, providing evidence that the Manifold Probe can discover manifolds which are causally involved in model behaviour.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 22 of its 35 references that are in the map. First posted on arXiv 18 May 2026.

Builds on

22 of 35 references.

References · 22

  1. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  2. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,359
  3. 2016Understanding intermediate layers using linear classifier probescited by 1,279
  4. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
  5. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,009
  6. 2022Toy Models of Superpositioncited by 928
  7. 2023Steering Language Models With Activation Engineeringcited by 824
  8. 2023Progress measures for grokking via mechanistic interpretabilitycited by 822
  9. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 673
  10. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 415
  11. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Modelscited by 313
  12. 2023Language Models Represent Space and Timecited by 306
10 more
  1. 2024Not All Language Model Features Are One-Dimensionally Linearcited by 201
  2. 2023The Clock and the Pizza: Two Stories in Mechanistic Explanation of Neural Networkscited by 153
  3. 2026Emotion Concepts and their Function in a Large Language Modelcited by 48
  4. 2025The Origins of Representation Manifolds in Large Language Modelscited by 46
  5. 2026When Models Manipulate Manifolds: The Geometry of a Counting Taskcited by 43
  6. 2026Symmetry in language statistics shapes the geometry of model representationscited by 20
  7. 2020Vintage Factor Analysis with Varimax Performs Statistical Inferencecited by 16
  8. 2026Valence-Arousal Subspace in LLMs: Circular Emotion Geometry and Multi-Behavioral Controlcited by 12
  9. 2026The Geometry of Representational Failures in Vision Language Modelscited by 5
  10. 2026Latent Structure of Affective Representations in Large Language Modelscited by 3

Built on by

Not cited yet.