Video Large Language Models (VLLMs) excel in video understanding, but their excessive visual tokens pose a significant computational challenge for real-world applications. Current methods aim to enhance inference efficiency by visual token pruning. However, they do not consider the dynamic characteristics and temporal dependencies of video frames, as they perceive video understanding as a multi-frame task. To address these challenges, we propose MMG-Vid, a novel training-free visual token pruning framework that removes redundancy by Maximizing Marginal Gains at both segment-level and token-level. Specifically, we first divide the video into segments based on frame similarity, and then dynamically allocate the token budget for each segment to maximize the marginal gain of each segment. Subsequently, we propose a temporal-guided DPC algorithm that jointly models inter-frame uniqueness and intra-frame diversity, thereby maximizing the marginal gain of each token. By combining both stages, MMG-Vid can maximize the utilization of the limited token budget, significantly improving efficiency while maintaining strong performance. Extensive experiments demonstrate that MMG-Vid can maintain over 99.5% of the original performance, while effectively reducing 75% visual tokens and accelerating the prefilling stage by 3.9x on LLaVA-OneVision-7B.

Sashuai Zhou, Qiang Zhou, Jijin Hu, Hanqing Yang, Yue Cao, Junpeng Ma, Yinchao Ma, Jun Song, Tiezheng Ge, Cheng Yu, Bo Zheng, Zhou Zhao. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Video large language models (VideoLLM) excel at video understanding, but face efficiency challenges due to the quadratic complexity of abundant visual tokens.Our systematic analysis of token compression methods for Vide-oLLMs reveals two critical issues: (i) overlooking distinctive visual signals across frames, leading to information loss; (ii) suffering from implementation constraints, causing incompatibility with modern architectures or efficient operators.To address these challenges, we distill three design principles for VideoLLM token compression and propose a plug-andplay inference acceleration framework "Video Compression Commander" (VidCom 2 ).By quantifying each frame's uniqueness, VidCom 2 adaptively adjusts compression intensity across frames, effectively preserving essential information while reducing redundancy in video sequences.Extensive experiments across various VideoLLMs and benchmarks demonstrate the superior performance and efficiency of our VidCom 2 .With only 25% visual tokens, VidCom 2 achieves 99.6% of the original performance on LLaVA-OV while reducing 70.8% of the LLM generation latency.Notably, our Frame Compression Adjustment strategy is compatible with other token compression methods to further improve their performance.Our code is available at https: //github.com/xuyang-liu16/VidCom2.

In vision-language models (VLMs), visual tokens usually consume a significant amount of computational overhead, despite their sparser information density compared to text tokens. To address this, most existing methods learn a network to prune redundant visual tokens and require additional training data. Differently, we propose an efficient training-free token optimization mechanism dubbed SparseVLM without extra parameters or fine-tuning costs. Concretely, given that visual tokens complement text tokens in VLMs for linguistic reasoning, we select visual-relevant text tokens to rate the significance of vision tokens within the self-attention matrix extracted from the VLMs. Then we progressively prune irrelevant tokens. To maximize sparsity while retaining essential information, we introduce a rank-based strategy to adaptively determine the sparsification ratio for each layer, alongside a token recycling method that compresses pruned tokens into more compact representations. Experimental results show that our SparseVLM improves the efficiency of various VLMs across a range of image and video understanding tasks. In particular, when LLaVA is equipped with SparseVLM, it achieves a 54% reduction in FLOPs, lowers CUDA time by 37%, and maintains an accuracy rate of 97%. Our code is available at https://github.com/Gumpest/SparseVLMs.

Sharding technology has become an important direction in blockchain scalability research. However, unreasonable account allocation during the sharding process has also brought about several challenges. First, the imbalance in transaction workloads between shards affects transaction efficiency. Shards with excessive workload can lead to congestion. Second, the high ratio of cross-shard transactions increases the communication overhead of the blockchain. In this paper, we propose FennelChain, a blockchain sharding protocol based on a graph partitioning algorithm, which allocates accounts in a rational manner to improve blockchain performance. We model the blockchain system using graph theory, mapping accounts and transactions in the system to vertices and edges in a graph, thus constructing a blockchain transaction graph. Furthermore, we propose the P-Fennel dynamic account partitioning algorithm, which reassigns accounts in the blockchain transaction graph to reduce the ratio of cross-shard transactions and address the problem of transaction load imbalance between shards. Finally, we conduct extensive experiments, and the results show that the proposed protocol outperforms other baselines in terms of throughput, transaction confirmation latency, and the ratio of cross-shard transactions.

The burgeoning field of Multimodal Large Language Models (MLLMs) has exhibited remarkable performance in diverse tasks such as captioning, commonsense reasoning, and visual scene understanding. However, the deployment of these large-scale MLLMs on client devices is hindered by their extensive model parameters, leading to a notable de-cline in generalization capabilities when these models are compressed for device deployment. Addressing this chal-lenge, we introduce a Cloud-Device Collaborative Contin-ual Adaptation framework, designed to enhance the performance of compressed, device-deployed MLLMs by lever-aging the robust capabilities of cloud-based, larger-scale MLLMs. Our framework is structured into three key components: a device-to-cloud uplink for efficient data transmission, cloud-based knowledge adaptation, and an optimized cloud-to-device downlink for model deployment. In the up-link phase, we employ an Uncertainty-guided Token Sam-pling (UTS) strategy to effectively filter out-of-distribution tokens, thereby reducing transmission costs and improving training efficiency. On the cloud side, we propose Adapter-based Knowledge Distillation (AKD) method to transfer refined knowledge from large-scale to compressed, pocket-size MLLMs. Furthermore, we propose a Dynamic Weight update Compression (DWC) strategy for the down-link, which adaptively selects and quantizes updated weight parameters, enhancing transmission efficiency and reducing the representational disparity between cloud and de-vice models. Extensive experiments on several multimodal benchmarks demonstrate the superiority of our proposed framework over prior Knowledge Distillation and device-cloud collaboration methods. Notably, we also validate the feasibility of our approach to real-world experiments.