Stateful Token Reduction for Long-Video Hybrid VLMs

• URL: http://arxiv.org/abs/2603.00198v1
• Date: Fri, 27 Feb 2026 08:11:06 GMT
• Title: Stateful Token Reduction for Long-Video Hybrid VLMs
• Authors: Jindong Jiang, Amala Sanjay Deshmukh, Kateryna Chumachenko, Karan Sapra, Zhiding Yu, Guilin Liu, Andrew Tao, Pavlo Molchanov, Jan Kautz, Wonmin Byeon,
• Abstract summary: We study query-conditioned token reduction for hybrid video vision-language models (VLMs)<n>We propose a low-to-high progressive reduction schedule and a unified language-aware scoring mechanism for both attention and Mamba blocks.<n>Under an aggressive compression setting, our approach delivers substantial prefilling speedups with near-baseline accuracy at test time.
• Score: 69.6930118088911
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Token reduction is an effective way to accelerate long-video vision-language models (VLMs), but most existing methods are designed for dense Transformers and do not directly account for hybrid architectures that interleave attention with linear-time state-space blocks (e.g., Mamba). We study query-conditioned token reduction for hybrid video VLMs and analyze reduction behavior through two properties: layerwise sparsity (how many tokens capture query-relevant information) and importance stability (whether token-importance rankings persist across depth). Although token importance is sparse within each layer, the set of important tokens changes across layers, so aggressive early pruning is unreliable. Motivated by this, we propose a low-to-high progressive reduction schedule and a unified language-aware scoring mechanism for both attention and Mamba blocks (using an implicit-attention proxy for Mamba), enabling all-layer token reduction in hybrids. Under an aggressive compression setting (retaining 25% of visual tokens), our approach delivers substantial prefilling speedups (3.8--4.2x) with near-baseline accuracy at test time, and light finetuning under reduction further improves performance on long-context video benchmarks.

Related papers

• Q Cache: Visual Attention is Valuable in Less than Half of Decode Layers for Multimodal Large Language Model [21.206033754351786]
  Multimodal large language models (MLLMs) are plagued by exorbitant inference costs attributable to the profusion of visual tokens.<n>Existing approaches focus on token-wise optimization, leveraging diverse token pruning techniques to eliminate non-crucial visual tokens.<n>We propose Lazy Attention, an efficient attention mechanism that enables cross-layer sharing of similar attention patterns.
  arXiv  Detail & Related papers  (2026-02-02T10:08:00Z)
• Fast SAM2 with Text-Driven Token Pruning [52.8350457627401]
  Segment Anything Model 2 (SAM2), a vision computation model has significantly advanced in prompt-driven video object segmentation.<n>SAM2 pipelines propagate all visual tokens produced by the image encoder through downstream temporal reasoning modules, regardless of their relevance to the target object.<n>We introduce a text-guided token pruning framework that improves inference efficiency by selectively reducing token density prior to temporal propagation.
  arXiv  Detail & Related papers  (2025-12-24T18:59:05Z)
• StreamingAssistant: Efficient Visual Token Pruning for Accelerating Online Video Understanding [29.539015046656615]
  We propose token pruning as a means to reduce context length while retaining critical information.<n>Specifically, we introduce a novel redundancy metric, Maximum Similarity to Spatially Adjacent Video Tokens (MSSAVT)<n>We also design a masked pruning strategy that ensures only mutually unadjacent tokens are pruned.
  arXiv  Detail & Related papers  (2025-12-14T05:35:11Z)
• Efficient Video Sampling: Pruning Temporally Redundant Tokens for Faster VLM Inference [5.146388234814547]
  Long videos often exceed the token budget of modern language models, leading to severe context limitations and latency issues.<n>We introduce Efficient Video Sampling (EVS), a simple, plug-and-play method for reducing token redundancy in videos by identifying and pruning temporally static patches.<n>EVS substantially reduces token count while maintaining semantic fidelity, enabling faster inference and longer input sequences.
  arXiv  Detail & Related papers  (2025-10-16T12:34:38Z)
• OjaKV: Context-Aware Online Low-Rank KV Cache Compression with Oja's Rule [54.37983890753086]
  We introduce OjaKV, a framework that integrates a strategic hybrid storage policy with online subspace adaptation.<n>OjaKV preserves crucial first and most recent tokens in full-rank, maintaining high-fidelity anchors for attention.<n>It applies low-rank compression by incrementally adapting the projection basis using Oja's algorithm for online principal component analysis.
  arXiv  Detail & Related papers  (2025-09-25T21:42:27Z)
• Dense Video Understanding with Gated Residual Tokenization [49.17263029080152]
  High temporal resolution is essential for capturing fine-grained details in video understanding.<n>Current benchmarks rely mostly on low-frame-rate sampling.<n>Dense Video Understanding (DVU) enables high-FPS video comprehension by reducing both tokenization time and token overhead.
  arXiv  Detail & Related papers  (2025-09-17T17:34:40Z)
• METok: Multi-Stage Event-based Token Compression for Efficient Long Video Understanding [55.38256656122857]
  We propose METok, a training-free, Multi-stage Event-based Token compression framework.<n>We show METok achieves an optimal trade-off between efficiency and accuracy by dynamically selecting informative visual tokens.<n>For instance, equipping LongVA-7B with METok realizes an 80.6% FLOPs reduction and 93.5% KV Cache memory savings.
  arXiv  Detail & Related papers  (2025-06-03T13:19:41Z)
• Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding [51.711605076319216]
  Diffusion-based large language models (Diffusion LLMs) have shown promise for non-autoregressive text generation with parallel decoding capabilities.<n>We introduce a novel block-wise approximate KV Cache mechanism tailored for bidirectional diffusion models, enabling cache reuse with negligible performance drop.<n>We propose a confidence-aware parallel decoding strategy that selectively decodes tokens exceeding a confidence threshold, mitigating dependency violations and maintaining generation quality.
  arXiv  Detail & Related papers  (2025-05-28T17:39:15Z)
• Skip-Vision: Efficient and Scalable Acceleration of Vision-Language Models via Adaptive Token Skipping [13.846838416902575]
  A key bottleneck stems from the proliferation of visual tokens required for fine-grained image understanding.<n>We propose Skip-Vision, a unified framework addressing both training and inference inefficiencies in vision-language models.<n> Experimental results demonstrate that Skip-Vision reduces training time by up to 35%, inference FLOPs by 75%, and latency by 45%.
  arXiv  Detail & Related papers  (2025-03-26T04:16:48Z)
• VQToken: Neural Discrete Token Representation Learning for Extreme Token Reduction in Video Large Language Models [35.38573641029626]
  We introduce the novel task of Extreme Short Token Reduction, which aims to represent entire videos using a minimal set of discrete tokens.<n>On the Extreme Short Token Reduction task, our VQToken compresses sequences to just 0.07 percent of their original length while incurring only a 0.66 percent drop in accuracy on the NextQA-MC benchmark.
  arXiv  Detail & Related papers  (2025-03-21T09:46:31Z)
• ZipVL: Efficient Large Vision-Language Models with Dynamic Token Sparsification [29.163757099307553]
  The efficiency of large vision-language models (LVLMs) is constrained by the computational bottleneck of the attention mechanism during the prefill phase.<n>We present ZipVL, an efficient inference framework designed for LVLMs through a dynamic ratio allocation strategy of important tokens.
  arXiv  Detail & Related papers  (2024-10-11T07:24:21Z)

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.