Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models

TL;DR

This paper demonstrates that textual steering vectors, derived from language models, can significantly improve the visual understanding capabilities of multimodal large language models across various tasks and architectures.

Contribution

It introduces a novel approach to steer multimodal models using text-derived vectors, filling a gap in existing techniques for MLLMs.

Findings

Text-derived steering improves multimodal accuracy.

Mean shift enhances spatial and counting tasks.

Method outperforms prompting and generalizes well.

Abstract

Steering methods have emerged as effective and targeted tools for guiding large language models' (LLMs) behavior without modifying their parameters. Multimodal large language models (MLLMs), however, do not currently enjoy the same suite of techniques, due in part to their recency and architectural diversity. Inspired by this gap, we investigate whether MLLMs can be steered using vectors derived from their text-only LLM backbone, via sparse autoencoders (SAEs), mean shift, and linear probing. We find that text-derived steering consistently enhances multimodal accuracy across diverse MLLM architectures and visual tasks. In particular, mean shift boosts spatial relationship accuracy on CV-Bench by up to +7.3% and counting accuracy by up to +3.3%, outperforming prompting and exhibiting strong generalization to out-of-distribution datasets. These results highlight textual steering vectors as a powerful, efficient mechanism for enhancing grounding in MLLMs with minimal additional data collection and computational overhead.