Imagen de Portada

Steering CLIP's vision transformer with sparse autoencoders

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Joseph, Sonia, Suresh, Praneet, Goldfarb, Ethan, Hufe, Lorenz, Gandelsman, Yossi, Graham, Robert, Bzdok, Danilo, Samek, Wojciech, Richards, Blake Aaron
Formato:	Preprint
Publicado:	2025
Materias:	Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Prisma: An Open Source Toolkit for Mechanistic Interpretability in Vision and Video
por: Joseph, Sonia, et al.
Publicado: (2025)

From Noise to Narrative: Tracing the Origins of Hallucinations in Transformers
por: Suresh, Praneet, et al.
Publicado: (2025)

Dyslexify: A Mechanistic Defense Against Typographic Attacks in CLIP
por: Hufe, Lorenz, et al.
Publicado: (2025)

From What to How: Attributing CLIP's Latent Components Reveals Unexpected Semantic Reliance
por: Dreyer, Maximilian, et al.
Publicado: (2025)

Quantifying LLM Attention-Head Stability: Implications for Circuit Universality
por: Bali, Karan, et al.
Publicado: (2026)

Interpreting ResNet-based CLIP via Neuron-Attention Decomposition
por: Bu, Edmund, et al.
Publicado: (2025)

The Unreasonable Effectiveness of Text Embedding Interpolation for Continuous Image Steering
por: Ekin, Yigit, et al.
Publicado: (2026)

Interpreting the Second-Order Effects of Neurons in CLIP
por: Gandelsman, Yossi, et al.
Publicado: (2024)

Interpreting CLIP's Image Representation via Text-Based Decomposition
por: Gandelsman, Yossi, et al.
Publicado: (2023)

Quantifying and Enabling the Interpretability of CLIP-like Models
por: Madasu, Avinash, et al.
Publicado: (2024)

The Uncanny Valley: A Comprehensive Analysis of Diffusion Models
por: Ghanem, Karam, et al.
Publicado: (2024)

Estimating Unknown Population Sizes Using the Hypergeometric Distribution
por: Hodgson, Liam, et al.
Publicado: (2024)

Learning Video Representations without Natural Videos
por: Yu, Xueyang, et al.
Publicado: (2024)

From Attribution to Action: A Human-Centered Application of Activation Steering
por: Labarta, Tobias, et al.
Publicado: (2026)

Vision Transformers Don't Need Trained Registers
por: Jiang, Nick, et al.
Publicado: (2025)

Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations
por: Jiang, Nick, et al.
Publicado: (2024)

In-Context Representation Hijacking
por: Yona, Itay, et al.
Publicado: (2025)

Same Task, Different Circuits: Disentangling Modality-Specific Mechanisms in VLMs
por: Nikankin, Yaniv, et al.
Publicado: (2025)

Scaling and evaluating sparse autoencoders
por: Gao, Leo, et al.
Publicado: (2024)

The Case for Model Science: Verify, Explore, Steer, Refine
por: Biecek, Przemyslaw, et al.
Publicado: (2026)

Model Science: getting serious about verification, explanation and control of AI systems
por: Biecek, Przemyslaw, et al.
Publicado: (2025)

Position: Explain to Question not to Justify
por: Biecek, Przemyslaw, et al.
Publicado: (2024)

From Weights to Activations: Is Steering the Next Frontier of Adaptation?
por: Ostermann, Simon, et al.
Publicado: (2026)

The More You See in 2D, the More You Perceive in 3D
por: Han, Xinyang, et al.
Publicado: (2024)

LLMs can see and hear without any training
por: Ashutosh, Kumar, et al.
Publicado: (2025)

Interpreting the Repeated Token Phenomenon in Large Language Models
por: Yona, Itay, et al.
Publicado: (2025)

Jailbreaking Vision-Language Models Through the Visual Modality
por: Azulay, Aharon, et al.
Publicado: (2026)

Decomposing multimodal embedding spaces with group-sparse autoencoders
por: Kaushik, Chiraag, et al.
Publicado: (2026)

Understanding sparse autoencoder scaling in the presence of feature manifolds
por: Michaud, Eric J., et al.
Publicado: (2025)

Applying sparse autoencoders to unlearn knowledge in language models
por: Farrell, Eoin, et al.
Publicado: (2024)

Iterative Inference in a Chess-Playing Neural Network
por: Sandmann, Elias, et al.
Publicado: (2025)

Benchmarking Uncertainty and its Disentanglement in multi-label Chest X-Ray Classification
por: Baur, Simon, et al.
Publicado: (2025)

Learning biologically relevant features in a pathology foundation model using sparse autoencoders
por: Le, Nhat Minh, et al.
Publicado: (2024)

Designing a system of labor market statistics and information / Robert S. Goldfarb, Arvil V. Adams
por: Goldfarb, Robert S
Publicado: (1993)

Towards the AI Historian: Agentic Information Extraction from Primary Sources
por: Hufe, Lorenz, et al.
Publicado: (2026)

Mitochondria‐nucleus crosstalk characterizes Alzheimer's disease across 1,5 million brain cells
por: Chloé Savignac, et al.
Publicado: (2025)

Can sparse autoencoders be used to decompose and interpret steering vectors?
por: Mayne, Harry, et al.
Publicado: (2024)

Investigating task-specific prompts and sparse autoencoders for activation monitoring
por: Tillman, Henk, et al.
Publicado: (2025)

Teaching Humans Subtle Differences with DIFFusion
por: Chiquier, Mia, et al.
Publicado: (2025)

An Empirical Study of Autoregressive Pre-training from Videos
por: Rajasegaran, Jathushan, et al.
Publicado: (2025)