A Survey on Mechanistic Interpretability for Multi-Modal Foundation Models

TL;DR

This survey reviews interpretability methods for multimodal foundation models, highlighting differences from language models and identifying key research gaps to improve understanding and control of these complex systems.

Contribution

It provides a structured taxonomy of interpretability techniques for MMFMs and compares them with unimodal models, addressing a significant research gap.

Findings

Systematic review of MMFM interpretability methods

Comparison between unimodal and multimodal interpretability approaches

Identification of critical research gaps in MMFM interpretability

Abstract

The rise of foundation models has transformed machine learning research, prompting efforts to uncover their inner workings and develop more efficient and reliable applications for better control. While significant progress has been made in interpreting Large Language Models (LLMs), multimodal foundation models (MMFMs) - such as contrastive vision-language models, generative vision-language models, and text-to-image models - pose unique interpretability challenges beyond unimodal frameworks. Despite initial studies, a substantial gap remains between the interpretability of LLMs and MMFMs. This survey explores two key aspects: (1) the adaptation of LLM interpretability methods to multimodal models and (2) understanding the mechanistic differences between unimodal language models and crossmodal systems. By systematically reviewing current MMFM analysis techniques, we propose a structured taxonomy of interpretability methods, compare insights across unimodal and multimodal architectures, and highlight critical research gaps.