English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

MMDiff: Multimodal Model Diffing for Feature Discovery and Control in MLLMs

Forum topic · 小凯 · 2026-08-11

Summary

MMDiff is a multimodal model-diffing framework introduced by researchers including Hunar Batra, Lachin Naghashyar, and Ashkan Khakzar (arXiv:2508.03805) that turns sparse autoencoders (SAEs) into feature-level interfaces for discovering and controlling multimodal behavior in Multimodal Large Language Models (MLLMs). The framework supports three uses: feature isolation by diffing a base-LM SAE against its multimodal-adapted counterpart, task-specific feature detection via per-token contrastive activation analysis, and feature-level control through causal ablation or steering of discovered feature directions. The authors trained multimodal SAEs for three MLLM families (LLaVA-MORE, PaliGemma 2, and InternVL3.5) and evaluated on visual spatial understanding, multimodal safety, and OCR. Removing discovered features selectively degraded target behavior by 12% on spatial tasks, 17% on OCR, and reduced attack success rate by 24% on multimodal safety attacks without affecting VQA performance. Steering these features improved spatial and OCR accuracy by +3.6% and +1.8% on average over standard single-layer steering baselines, showing multimodal SAEs can serve as mechanisms for auditing and controlling MLLM behavior.

Paper Overview

Field: NLP Authors: Hunar Batra, Lachin Naghashyar, Ashkan Khakzar arXiv: 2508.03805

Abstract

Multimodal Large Language Models (MLLMs) exhibit strong visual understanding, yet the internal features that cause these behaviors remain difficult to identify, audit, or control. While applicable to post-hoc inspection, hidden states decomposed into interpretable feature directions using sparse autoencoders (SAEs) neither readily isolate which features are changed by multimodal training, nor are they directly useful for targeted control. The authors introduce MMDiff, a multimodal model-diffing framework that trains multimodal SAEs and turns them into feature-level interfaces for discovering and controlling multimodal behavior.

Key Capabilities

MMDiff supports three uses:

1. Feature isolation — diffing a base-LM SAE against its multimodal-adapted counterpart to identify features altered by multimodal training. 2. Task-specific feature detection — per-token contrastive activation analysis to isolate causal features. 3. Feature-level control — causal ablation or steering of the discovered feature directions.

Experiments and Results

The authors trained multimodal SAEs for three MLLM families — LLaVA-MORE, PaliGemma 2, and InternVL3.5 — and evaluated on visual spatial understanding, multimodal safety, and OCR.

  • Removing discovered features selectively degraded target behavior by 12% on spatial tasks and 17% on OCR, and reduced attack success rate by 24% on multimodal safety attacks, without affecting VQA performance.
  • Steering these features improved spatial and OCR accuracy by +3.6% and +1.8% on average over standard single-layer steering baselines.

Conclusion

These results show that multimodal SAEs can serve not only as interpretability tools but also as mechanisms for auditing, steering, and controlling MLLM behavior toward safer and more capable generation.

Tags

#mllm#sparse-autoencoders#interpretability#model-diffing#multimodal-safety#feature-steering#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178633335