English(EN) SheafPrune: Sheaf-Guided Graph Propagation for Visual Token Pruning

SheafPrune 通过修剪视觉标记来降低多模态大语言模型的成本

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-12 04:00

研究人员开发了 SheafPrune，一种通过修剪不相关的视觉标记来降低多模态大语言模型计算成本的新颖方法。该技术利用受束几何学启发的基于图的方法在文本和视觉标记之间传播相关性分数，并校正节点度以提高准确性。SheafPrune 拥有约 450 万个参数，与 LLaVA-NeXT-8B 主干网络进行端到端训练，并可适应各种标记预算。实验表明，SheafPrune 在保持多个基准测试的高准确性的同时，显著加快了预填充时间并降低了 FLOPs。 AI

影响降低了多模态大语言模型的推理成本，可能支持更广泛的部署和更快的处理。

排序理由详细介绍优化大语言模型推理新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CV 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: 8 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍优化大语言模型推理新方法的学术论文。[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: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.CV TIER_1 English(EN) · Xiaoyang Ma, Chi Wang, Hong Sun · 2026-10-12 04:00

SheafPrune: 用于视觉令牌剪枝的束引导图传播

arXiv:2610.13263v1 Announce Type: new Abstract: Multimodal large language models process the complete visual token sequence, increasing prefill cost even when many tokens are irrelevant to the instruction. Token-wise saliency captures limited inter-token dependence; graph-based s…

本文如何被排名

报道来源 [1]

SheafPrune: 用于视觉令牌剪枝的束引导图传播

相关实体

相关话题