Language Models Can Explain Visual Features via Steering

TL;DR

This paper introduces a novel method called Steering that uses causal interventions in vision-language models to generate explanations for features in vision models, improving interpretability and scalability.

Contribution

The paper presents a new approach leveraging causal interventions and language models to explain vision features, outperforming traditional correlation-based methods.

Findings

Explanation quality improves with larger language models.

Steering-informed Top-k achieves state-of-the-art results.

Method provides scalable and automated interpretability.

Abstract

Sparse Autoencoders uncover thousands of features in vision models, yet explaining these features without requiring human intervention remains an open challenge. While previous work has proposed generating correlation-based explanations based on top activating input examples, we present a fundamentally different alternative based on causal interventions. We leverage the structure of Vision-Language Models and steer individual SAE features in the vision encoder after providing an empty image. Then, we prompt the language model to explain what it ``sees'', effectively eliciting the visual concept represented by each feature. Results show that Steering offers an scalable alternative that complements traditional approaches based on input examples, serving as a new axis for automated interpretability in vision models. Moreover, the quality of explanations improves consistently with the scale of the language model, highlighting our method as a promising direction for future research. Finally, we propose Steering-informed Top-k, a hybrid approach that combines the strengths of causal interventions and input-based approaches to achieve state-of-the-art explanation quality without additional computational cost.