Lune

EMNLP Findings2025

Unveiling Multimodal Processing: Exploring Activation Patterns in Multimodal LLMs for Interpretability and Efficiency

Chuan Wu, Meng Su, Youxuan Fang, Shaolin Zhu

Abstract

Recent Multimodal Large Language Models (MLLMs) have achieved remarkable advancements, yet their internal mechanisms for concurrently processing diverse modalities like text, image, and audio remain largely opaque. In this paper, we propose a methodology to convert dense MLLMs into fine-grained Mixtureof-Experts (MoE) architectures. This allows us to visually investigate their multimodal activation patterns through expert activation frequency heatmaps. Conducting comprehensive experiments on representative MLLMs, we analyze the similarities and differences in internal neuron activations when handling distinct modalities. Specifically, we examine the distribution of high-frequency activated experts, the distinct roles of high-frequency (e.g., fundamental logic) and low-frequency (e.g., domainspecific concepts) multimodal shared experts, and the prevalence and localization of modalityspecific experts. Furthermore, we explore leveraging these discovered activation discrepancies to guide sparse activation and model pruning. Experimental results demonstrate that our approach substantially outperforms random expert pruning and can achieve comparable or even superior performance to the original unpruned models while utilizing significantly fewer active parameters. Our work not only sheds light on the multimodal processing mechanisms within MLLMs but also provides a practical pathway toward developing more interpretable and efficient multimodal systems.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

1 top-tier paper has cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • The Revolution of Multimodal Large Language Models: A Survey cites Evaluating Object Hallucination in Large Vision-Language Models
  • MoEfication: Transformer Feed-forward Layers are Mixtures of Experts cites Locating and Editing Factual Associations in GPT
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers1

  1. MACS: Modality-Aware Capacity Scaling for Efficient Multimodal MoE InferenceBo Li, Chuan Wu, Shaolin ZhuACL 2026

Builds on12

  1. Locating and Editing Factual Associations in GPTKevin Meng, David Bau, Alex Andonian, Yonatan BelinkovNeurIPS 2022 · 3,415 citations
  2. Are We on the Right Way for Evaluating Large Vision-Language Models?Lin Chen, Jinsong Li, Xiaoyi Dong, Pan Zhang et al.NeurIPS 2024 · 1,029 citations
  3. Evaluating Object Hallucination in Large Vision-Language ModelsYifan Li, Yifan Du, Kun Zhou, Jinpeng Wang et al.EMNLP 2023 · 344 citations
  4. The Revolution of Multimodal Large Language Models: A SurveyDavide Caffagni, Federico Cocchi, Luca Barsellotti, Nicholas Moratelli et al.ACL Findings 2024 · 99 citations
  5. MoEfication: Transformer Feed-forward Layers are Mixtures of ExpertsZhengyan Zhang, Yankai Lin, Zhiyuan Liu, Peng Li et al.ACL Findings 2022 · 48 citations

Related papers

  1. Unraveling Babel: Exploring Multilingual Activation Patterns of LLMs and Their ApplicationsWeize Liu, Yinlong Xu, Hongxia Xu, Jintai Chen et al.EMNLP 2024 · 1 citation
  2. C-GNN-PRUNE: A Unified Graph-Based Framework for Structure-Aware Pruning of Mixture-of-Experts ModelsLin Li, Yan Wang, Zhuopeng WangAAAI 2026 · 1 citation
  3. The Expert Strikes Back: Interpreting Mixture-of-Experts Language Models at Expert LevelJeremy Herbst, Stefan Wermter, Jae Hee LeeICML 2026 · 9 citations
  4. FastMMoE: Accelerating Multimodal Large Language Models through Dynamic Expert Activation and Routing-Aware Token PruningGuoyang Xia, Yifeng Ding, Fengfa Li, Lei Ren, Wei Chen, Fangxiang Feng, Xiaojie WangCVPR Findings 2026
  5. Multimodal Contrastive Learning with LIMoE: the Language-Image Mixture of ExpertsBasil Mustafa, Carlos Riquelme, Joan Puigcerver, Rodolphe Jenatton et al.NeurIPS 2022 · 359 citations
Dusk over the sea between two cliffs drawn in fine vertical lines