# Beyond Intermediate States: Explaining Visual Redundancy through Language

arXiv: https://arxiv.org/abs/2503.20540 · HTML: https://mlsanity.com/paper/2503.20540/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 44 of its 52 references that are in the map.
- First posted on arXiv 26 Mar 2025.

## Builds on (most cited first, 30 of 44)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering](https://mlsanity.com/paper/1612.00837/) (2016)
- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [Evaluating Object Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2305.10355/) (2023)
- [MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models](https://mlsanity.com/paper/2306.13394/) (2023)
- [MVBench: A Comprehensive Multi-modal Video Understanding Benchmark](https://mlsanity.com/paper/2311.17005/) (2023)
- [Deep Visual-Semantic Alignments for Generating Image Descriptions](https://mlsanity.com/paper/1412.2306/) (2014)
- [SEED-Bench: Benchmarking Multimodal LLMs with Generative Comprehension](https://mlsanity.com/paper/2307.16125/) (2023)
- [Vision Transformers Need Registers](https://mlsanity.com/paper/2309.16588/) (2023)
- [SPICE: Semantic Propositional Image Caption Evaluation](https://mlsanity.com/paper/1607.08822/) (2016)
- [An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models](https://mlsanity.com/paper/2403.06764/) (2024)
- [nocaps: novel object captioning at scale](https://mlsanity.com/paper/1812.08658/) (2018)
- [LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models](https://mlsanity.com/paper/2403.15388/) (2024)
- [SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference](https://mlsanity.com/paper/2410.04417/) (2024)
- [LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding](https://mlsanity.com/paper/2410.17434/) (2024)
- [VisionZip: Longer is Better but Not Necessary in Vision Language Models](https://mlsanity.com/paper/2412.04467/) (2024)
- [PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction](https://mlsanity.com/paper/2410.17247/) (2024)
- [MANTIS: Interleaved Multi-Image Instruction Tuning](https://mlsanity.com/paper/2405.01483/) (2024)
- [MuirBench: A Comprehensive Benchmark for Robust Multi-image Understanding](https://mlsanity.com/paper/2406.09411/) (2024)
- [TokenPacker: Efficient Visual Projector for Multimodal LLM](https://mlsanity.com/paper/2407.02392/) (2024)
- [Ovis: Structural Embedding Alignment for Multimodal Large Language Model](https://mlsanity.com/paper/2405.20797/) (2024)
- [Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolution](https://mlsanity.com/paper/2409.12961/) (2024)
- [Beyond Text-Visual Attention: Exploiting Visual Cues for Effective Token Pruning in VLMs](https://mlsanity.com/paper/2412.01818/) (2024)
- [Fit and Prune: Fast and Training-free Visual Token Pruning for Multi-modal Large Language Models](https://mlsanity.com/paper/2409.10197/) (2024)

## Built on by (most cited first, 7 of 7)

- [A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models](https://mlsanity.com/paper/2511.15098/) (2025)
- [M3-ACE: Rectifying Visual Perception in Multimodal Math Reasoning via Multi-Agentic Context Engineering](https://mlsanity.com/paper/2603.08369/) (2026)
- [Visual-Redundancy-Controlled Parallel Decoding for Diffusion-Based Multimodal Large Language Models](https://mlsanity.com/paper/2605.25820/) (2026)
- [Interpret, prune and distill Donut : towards lightweight VLMs for VQA on document](https://mlsanity.com/paper/2509.26235/) (2025)
- [A Cross-Modal Prompt Injection Attack against Large Vision-Language Models with Image-Only Perturbation](https://mlsanity.com/paper/2605.16090/) (2026)
- [Clustering and Token Denoising for Faster and More Robust VLMs](https://mlsanity.com/paper/2608.19285/) (2026)
- [From Saliency to Discriminability: Rank-Preserving Visual Token Pruning for VLM Rerankers](https://mlsanity.com/paper/2609.00667/) (2026)

## Abstract (arXiv)

Multi-modal Large Langue Models (MLLMs) often process thousands of visual tokens, which consume a significant portion of the context window and impose a substantial computational burden. Prior work has empirically explored visual token pruning methods based on MLLMs' intermediate states (e.g., attention scores). However, they have limitations in precisely defining visual redundancy due to their inability to capture the influence of visual tokens on MLLMs' visual understanding (i.e., the predicted probabilities for textual token candidates). To address this issue, we manipulate the visual input and investigate variations in the textual output from both token-centric and context-centric perspectives, achieving intuitive and comprehensive analysis. Experimental results reveal that visual tokens with low ViT-[cls] association and low text-to-image attention scores can contain recognizable information and significantly contribute to images' overall information. To develop a more reliable method for identifying and pruning redundant visual tokens, we integrate these two perspectives and introduce a context-independent condition to identify redundant prototypes from training images, which probes the redundancy of each visual token during inference. Extensive experiments on single-image, multi-image and video comprehension tasks demonstrate the effectiveness of our method, notably achieving 90% to 110% of the performance while pruning 80% to 90% of visual tokens.