ViCA: Efficient Multimodal LLMs with Vision-Only Cross-Attention

• URL: http://arxiv.org/abs/2602.07574v1
• Date: Sat, 07 Feb 2026 14:46:05 GMT
• Title: ViCA: Efficient Multimodal LLMs with Vision-Only Cross-Attention
• Authors: Wenjie Liu, Hao Wu, Xin Qiu, Yingqi Fan, Yihan Zhang, Anhao Zhao, Yunpu Ma, Xiaoyu Shen,
• Abstract summary: ViCA is a minimal MLLM architecture in which visual tokens bypass all self-attention and feed-forward layers, interacting with text solely through sparse cross-attention at selected layers.<n>ViCA preserves 98% of baseline accuracy while reducing visual-side to 4%, consistently achieving superior performance-efficiency trade-offs.
• Score: 22.397648349603696
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Modern multimodal large language models (MLLMs) adopt a unified self-attention design that processes visual and textual tokens at every Transformer layer, incurring substantial computational overhead. In this work, we revisit the necessity of such dense visual processing and show that projected visual embeddings are already well-aligned with the language space, while effective vision-language interaction occurs in only a small subset of layers. Based on these insights, we propose ViCA (Vision-only Cross-Attention), a minimal MLLM architecture in which visual tokens bypass all self-attention and feed-forward layers, interacting with text solely through sparse cross-attention at selected layers. Extensive evaluations across three MLLM backbones, nine multimodal benchmarks, and 26 pruning-based baselines show that ViCA preserves 98% of baseline accuracy while reducing visual-side computation to 4%, consistently achieving superior performance-efficiency trade-offs. Moreover, ViCA provides a regular, hardware-friendly inference pipeline that yields over 3.5x speedup in single-batch inference and over 10x speedup in multi-batch inference, reducing visual grounding to near-zero overhead compared with text-only LLMs. It is also orthogonal to token pruning methods and can be seamlessly combined for further efficiency gains. Our code is available at https://github.com/EIT-NLP/ViCA.

Related papers

• Nüwa: Mending the Spatial Integrity Torn by VLM Token Pruning [82.39668822222386]
  Vision token pruning has proven to be an effective acceleration technique for the efficient Vision Language Model (VLM)<n>We propose $textNwa$, a two-stage token pruning framework that enables efficient feature aggregation while maintaining spatial integrity.<n>Experiments demonstrate that $textNwa$ achieves SOTA performance on multiple VQA benchmarks (from 94% to 95%) and yields substantial improvements on visual grounding tasks (from 7% to 47%)
  arXiv  Detail & Related papers  (2026-02-03T00:51:03Z)
• From One-to-One to Many-to-Many: Dynamic Cross-Layer Injection for Deep Vision-Language Fusion [91.35078719566472]
  Vision-Language Models (VLMs) create a severe visual feature bottleneck by using a crude, asymmetric connection.<n>We introduce Cross-Layer Injection (CLI), a novel and lightweight framework that forges a dynamic many-to-many bridge between the two modalities.
  arXiv  Detail & Related papers  (2026-01-15T18:59:10Z)
• Parallel Vision Token Scheduling for Fast and Accurate Multimodal LMMs Inference [68.4758228017823]
  ParVTS partitions visual tokens into subject and non-subject groups, processes them in parallel to transfer their semantics into question tokens, and discards the non-subject path mid-inference.<n>Experiments show that ParVTS prunes up to 88.9% of visual tokens with minimal performance drop, achieving 1.77x speedup and 70% FLOPs reduction.
  arXiv  Detail & Related papers  (2025-11-24T08:29:36Z)
• $\mathcal{V}isi\mathcal{P}runer$: Decoding Discontinuous Cross-Modal Dynamics for Efficient Multimodal LLMs [26.779915891040236]
  We propose emphVisiPruner, a training-free pruning framework that reduces up to 99% of vision-related attention computations and 53.9% of FLOPs on LLaVA-v1.5 7B.<n>Our insights further provide actionable guidelines for training efficient MLLMs by aligning model architecture with its intrinsic layer-wise processing dynamics.
  arXiv  Detail & Related papers  (2025-10-20T06:40:17Z)
• CoViPAL: Layer-wise Contextualized Visual Token Pruning for Large Vision-Language Models [75.88232735646018]
  Large Vision-Language Models (LVLMs) process multimodal inputs consisting of text tokens and vision tokens extracted from images or videos.<n>Existing methods attempt to prune redundant vision tokens, revealing substantial redundancy in visual representations.<n>We propose CoViPAL, a layer-wise contextualized visual token pruning method that employs a Plug-and-Play Pruning Module (PPM) to predict and remove redundant vision tokens before they are processed by the LVLM.
  arXiv  Detail & Related papers  (2025-08-24T07:47:00Z)
• InternVL-X: Advancing and Accelerating InternVL Series with Efficient Visual Token Compression [1.8893427856534721]
  We propose InternVL-X, which outperforms the InternVL model in both performance and efficiency.<n>By utilizing 20% or fewer visual tokens, InternVL-X achieves state-of-the-art performance on 7 public MLLM benchmarks, and improves the average metric by 2.34% across 12 tasks.
  arXiv  Detail & Related papers  (2025-03-27T09:31:35Z)
• LVPruning: An Effective yet Simple Language-Guided Vision Token Pruning Approach for Multi-modal Large Language Models [18.489240454283834]
  We introduce Language-Guided Vision Token Pruning (LVPruning) for MLLMs.<n>LVPruning employs cross-attention modules to compute the importance of vision tokens based on their interaction with language tokens.<n>Experiments show that LVPruning can effectively reduce up to 90% of vision tokens by the middle layer of LLaVA-1.5.
  arXiv  Detail & Related papers  (2025-01-23T13:31:51Z)
• Accelerating Multimodal Large Language Models via Dynamic Visual-Token Exit and the Empirical Findings [66.04061083611863]
  Excessive use of visual tokens in existing Multimoal Large Language Models (MLLMs) often exhibits obvious redundancy and brings in prohibitively expensive computation.<n>We propose a simple yet effective method to improve the efficiency of MLLMs, termed dynamic visual-token exit (DyVTE)<n>DyVTE uses lightweight hyper-networks to perceive the text token status and decide the removal of all visual tokens after a certain layer.
  arXiv  Detail & Related papers  (2024-11-29T11:24:23Z)
• Sparsity Meets Similarity: Leveraging Long-Tail Distribution for Dynamic Optimized Token Representation in Multimodal Large Language Models [6.467840081978855]
  multimodal large language models (MM-LLMs) have achieved significant success in various tasks.<n>Main computational burden arises from processingd text and visual tokens.<n>We propose a dynamic pruning algorithm that identifies the inflection point in the visual CLS token similarity curve.
  arXiv  Detail & Related papers  (2024-09-02T10:49:10Z)
• Towards Semantic Equivalence of Tokenization in Multimodal LLM [149.11720372278273]
  Vision tokenization is essential for semantic alignment between vision and language.<n>This paper proposes a novel dynamic Semantic-Equivalent Vision Tokenizer (SeTok)<n>SeTok groups visual features into semantic units via a dynamic clustering algorithm.<n>The resulting vision tokens effectively preserve semantic integrity and capture both low-frequency and high-frequency visual features.
  arXiv  Detail & Related papers  (2024-06-07T17:55:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.