新的MWOP技术通过定向剪枝提升MLLM效率

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-02 04:00

研究人员开发了一种名为“模态感知宽度操作剪枝”（MWOP）的新方法，以提高多模态大语言模型（MLLM）的效率。MWOP通过独立剪枝视觉到视觉、文本到视觉和文本到文本的注意力路径，并分别选择视觉和文本输入的FFN通道，来解决注意力头和前馈网络（FFN）通道内的冗余问题。这种方法在保留令牌序列的同时减少了计算量，从而显著加快了速度。在LLaVA-OneVision-7B上，MWOP单独实现了1.6倍的预填充速度提升，且性能损失极小；与令牌压缩方法结合使用时，速度提升进一步增加。 AI

影响该方法可能导致多模态人工智能系统更高效的部署和更快的推理。

排序理由详细介绍模型效率新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 13 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍模型效率新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, infra; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Xudong Wang, Hao Wu, Haozhe Hu, Peiran Yin, Xinghao Chen, Yunpu Ma, Wei Zhang, Xiaoyu Shen · 2026-10-02 04:00

MWOP：面向高效MLLM的模态感知宽度操作剪枝

arXiv:2610.01434v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) incur substantial inference costs when processing long visual-textual sequences. While existing operation compression methods exploit modality-level redundancy, they largely treat computati…

本文如何被排名

报道来源 [1]

MWOP：面向高效MLLM的模态感知宽度操作剪枝

相关实体

相关话题