The paper introduces Visual Token Coding (VTC), a token compression method for video multimodal large language models that mimics classical video coding by predicting I/P frames and measuring residuals to reduce token redundancy. VTC is extended with dynamic features—Dynamic Resolution Input, Dynamic Token Allocation, and Spatial Coverage Top‑K—forming VTC_Dy, which can be applied to existing MLLMs without additional tuning. Experiments on three MLLMs and multiple video benchmarks show that VTC_Dy retains over 100% of average performance with a 50% token budget and 97.8% with a 25% budget, while the code is publicly available.
By Chenxin Fang, Tao Chen, JunChao You, Jun Peng, Yiyi Zhou, Rongrong Ji
arXiv:2603. 00198v2 Announce Type: replace-cross Abstract: Token reduction accelerates long-video vision--language models (VLMs), but existing methods target Transformers, where reduction is treated as token pruning.
By Jindong Jiang, Amala Sanjay Deshmukh, Kateryna Chumachenko, Karan Sapra, Zhiding Yu, Guilin Liu, Andrew Tao, Pavlo Molchanov, Jan Kautz, Wonmin Byeon
arXiv:2512. 08240v2 Announce Type: replace-cross Abstract: Vision-language models (VLMs) rely on hundreds of visual tokens, leading to high computational and memory costs.
By Jusheng Zhang, Xiaoyang Guo, Tongyu Mo, Qinhan Lv, Wenhao Chai, Jian Wang, Keze Wang, Liang Lin
In video understanding, vision-language models (VLMs) must ingest massive numbers of visual tokens, causing the computational and memory cost of the prefill stage to rise sharply. Such visual sequences are highly redundant along the spatio-temporal dimension, yet a high compression ratio is often accompanied by the loss of critical details.
arXiv:2608. 09176v1 Announce Type: cross Abstract: Visual token compression for vision--language models (VLMs) has largely relied on criteria such as attention, redundancy, and uncertainty to maximize average accuracy under a fixed compute budget, implicitly assuming that all errors carry equal cost.
By Jingbo Wen, Liang He, Mingyu Cao, Haoyu Wang, Minxuan Hu, Kangning Cui, Xilu Wang
arXiv:2606. 28551v1 Announce Type: cross Abstract: Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies.
By Matteo Farina, Vishaal Udandarao, Thao Nguyen, Selim Kuzucu, Maximilian B\"other, Andreas Hochlehnert, Adhiraj Ghosh, Marianna Nezhurina, Karsten Roth, Joschka Struber, Yuhui Zhang, Sebastian Dziadzio, Elaine Sui, Soumya Jahagirdar, Dhruba Ghosh, Hasan Hammoud, Thomas De Min, Simone Caldarella, Jehanzeb Mirza, Sedrick Keh, Mehdi Cherti, Hilde Kuehne, Bernt Schiele, Serena Yeung-Levy, Muhammad Ferjad Naeem, Federico Tombari, Ana Klimovic, Elisa Ricci, Matthias Bethge, Sewoong Oh, Ameya Prabhu, Alessio Tonioni, Jenia Jitsev, Massimiliano Mancini, Ludwig Schmidt, Nikhil Parthasarathy
PACE introduces a training‑free Condense‑and‑Extract framework that speeds up Vision‑Language Model inference by first adaptively downsampling visual inputs before encoding and then selectively retaining essential tokens during decoding. The Adaptive Pixel Compressor (APC) reduces encoder workload while preserving global context, and the Dynamic Dual‑Attention Extractor (DDAE) keeps task‑critical details by fusing visual and language signals. Applied to Qwen2.5‑VL‑7B, PACE maintains 93.8% of performance using only 10% of visual tokens, achieving a 3.1× speedup in time to first token.
By Junjie Liu, Shengyuan Ye, Xu Chen
VisCache introduces a two-stage, plug‑and‑play framework for pruning visual key‑value caches in Vision Large Language Models without retraining. The first stage filters out temporally redundant keyframes, while the second stage, PruneKV, applies a parabolic layer‑wise budget and asymmetric update to selectively prune keys and fuse values, preserving essential context. Experiments show up to 2.35× speedup and significant memory savings with only 19–28% of the original cache retained, outperforming existing baselines.
By Lyuke Wang, Zhuo Li, Guangxu Zhu
arXiv:2606. 07577v1 Announce Type: new Abstract: Audio-visual large language models (LLMs) hold strong promise for long-form video understanding, yet their long-video inference is fundamentally limited by the linear growth of video tokens and key-value (KV) caches.
By Guangzhi Sun, Yixuan Li, Yudong Yang, Chao Zhang
arXiv:2607. 28627v1 Announce Type: cross Abstract: Long visual context poses a challenge for vision-language models: performance degrades as the number of distractors grows, and processing all tokens at once is computationally infeasible under GPU memory constraints.
By Yao Xiao, Reuben Tan, Zhen Zhu, Yuqun Wu, Jianfeng Gao, Derek Hoiem
arXiv:2608. 03112v1 Announce Type: cross Abstract: Vision-language models excel at image and video understanding but suffer from high inference latency due to the need to process thousands of tokens per image, limiting their deployment on resource-constrained edge devices and in real-time surveillance applications.
By Paribesh Regmi, Qingshuang Chen, Chi Zhang, Heba Aly, Yelin Kim, Hongda Mao
arXiv:2608. 01263v1 Announce Type: new Abstract: On-policy distillation (OPD) samples trajectories from the current student policy and minimizes token-level divergence between student and teacher next-token distributions at prefixes along those trajectories.
By Leyan Xue, Feng Xiong, Mingjun Ma, Changqing Zhang