Image de couverture de livre

Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Wurgaft, Daniel, Rager, Can, Kowal, Matthew, Shyam, Vasudev, Feucht, Sheridan, Bhalla, Usha, Haklay, Tal, Bigelow, Eric, Sarfati, Raphael, McGrath, Thomas, Lewis, Owen, Merullo, Jack, Goodman, Noah, Fel, Thomas, Geiger, Atticus, Lubana, Ekdeep Singh
Format:	Preprint
Publié:	2026
Sujets:	Machine Learning
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Do Sparse Autoencoders Capture Concept Manifolds?
par: Bhalla, Usha, et autres
Publié: (2026)

Arithmetic in the Wild: Llama uses Base-10 Addition to Reason About Cyclic Concepts
par: Feucht, Sheridan, et autres
Publié: (2026)

Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space
par: Bigelow, Eric, et autres
Publié: (2026)

The Shape of Beliefs: Geometry, Dynamics, and Interventions along Representation Manifolds of Language Models' Posteriors
par: Sarfati, Raphaël, et autres
Publié: (2026)

Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
par: Zur, Amir, et autres
Publié: (2025)

Belief Dynamics Reveal the Dual Nature of In-Context Learning and Activation Steering
par: Bigelow, Eric, et autres
Publié: (2025)

Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought
par: Boppana, Siddharth, et autres
Publié: (2026)

Features as Rewards: Scalable Supervision for Open-Ended Tasks via Interpretability
par: Prasad, Aaditya Vikram, et autres
Publié: (2026)

Uncovering Conceptual Blindspots in Generative Image Models Using Sparse Autoencoders
par: Bohacek, Matyas, et autres
Publié: (2025)

Projecting Assumptions: The Duality Between Sparse Autoencoders and Concept Geometry
par: Hindupur, Sai Sumedh R., et autres
Publié: (2025)

From Flat to Hierarchical: Extracting Sparse Representations with Matching Pursuit
par: Costa, Valérie, et autres
Publié: (2025)

Evaluating Sparse Autoencoders: From Shallow Design to Matching Pursuit
par: Costa, Valérie, et autres
Publié: (2025)

In-Context Learning Strategies Emerge Rationally
par: Wurgaft, Daniel, et autres
Publié: (2025)

Towards Reliable Evaluation of Behavior Steering Interventions in LLMs
par: Pres, Itamar, et autres
Publié: (2024)

Priors in Time: Missing Inductive Biases for Language Model Interpretability
par: Lubana, Ekdeep Singh, et autres
Publié: (2025)

Archetypal SAE: Adaptive and Stable Dictionary Learning for Concept Extraction in Large Vision Models
par: Fel, Thomas, et autres
Publié: (2025)

Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry
par: Fel, Thomas, et autres
Publié: (2025)

In-Context Learning Dynamics with Random Binary Sequences
par: Bigelow, Eric J., et autres
Publié: (2023)

Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations
par: Geiger, Atticus, et autres
Publié: (2023)

Interpretability at Scale: Identifying Causal Mechanisms in Alpaca
par: Wu, Zhengxuan, et autres
Publié: (2023)

How Causal Abstraction Underpins Computational Explanation
par: Geiger, Atticus, et autres
Publié: (2025)

Activation Steering via Generative Causal Mediation
par: Sankaranarayanan, Aruna, et autres
Publié: (2026)

HyperSteer: Activation Steering at Scale with Hypernetworks
par: Sun, Jiuding, et autres
Publié: (2025)

How Do LLMs Persuade? Linear Probes Can Uncover Persuasion Dynamics in Multi-Turn Conversations
par: Jaipersaud, Brandon, et autres
Publié: (2025)

Abrupt Learning in Transformers: A Case Study on Matrix Completion
par: Gopalani, Pulkit, et autres
Publié: (2024)

Emergence of Hierarchical Emotion Organization in Large Language Models
par: Zhao, Bo, et autres
Publié: (2025)

Comparisons of the 1995 and 1998 coral bleaching events on the patch reefs of San Salvador Island, Bahamas
par: Thomas A. McGrath
Publié: (2003)

Evaluating Open-Source Sparse Autoencoders on Disentangling Factual Knowledge in GPT-2 Small
par: Chaudhary, Maheep, et autres
Publié: (2024)

Vector Arithmetic in Concept and Token Subspaces
par: Feucht, Sheridan, et autres
Publié: (2025)

Sparks of Explainability: Recent Advancements in Explaining Large Vision Models
par: Fel, Thomas
Publié: (2025)

Constructing Interpretable Features from Compositional Neuron Groups
par: Shafran, Or, et autres
Publié: (2025)

Analyzing (In)Abilities of SAEs via Formal Languages
par: Menon, Abhinav, et autres
Publié: (2024)

A Reply to Makelov et al. (2023)'s "Interpretability Illusion" Arguments
par: Wu, Zhengxuan, et autres
Publié: (2024)

Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment
par: Thasarathan, Harrish, et autres
Publié: (2025)

Why Larger Models Learn More: Effects of Capacity, Interference, and Rare-Task Retention
par: Huang, Jing, et autres
Publié: (2026)

A Percolation Model of Emergence: Analyzing Transformers Trained on a Formal Language
par: Lubana, Ekdeep Singh, et autres
Publié: (2024)

Competition Dynamics Shape Algorithmic Phases of In-Context Learning
par: Park, Core Francisco, et autres
Publié: (2024)

Compositional Abilities Emerge Multiplicatively: Exploring Diffusion Models on a Synthetic Task
par: Okawa, Maya, et autres
Publié: (2023)

How Do Transformers Learn Variable Binding in Symbolic Programs?
par: Wu, Yiwei, et autres
Publié: (2025)

InfoTrac's SearchBank Databases: Business Information and More.
par: Mehta, Usha, et autres
Publié: (1997)