PulseAugur
中
实时 20:40:02
English(EN) MixTok: Adaptive Visual Tokenization with Global 1D Tokens and Sparse 2D Refinement

新方法优化大型视觉语言模型的视觉令牌化 · 跟踪到 4 个来源

研究人员开发了几种新方法来优化大型视觉语言模型(VLM)的视觉令牌化,旨在降低计算开销同时保持准确性。MixTok 引入了一种混合方法,结合了全局一维令牌和用于局部精炼的稀疏二维令牌,在 ImageNet 上取得了有竞争力的性能。ERASE 采用两阶段框架,首先去除图像级冗余,然后修剪与指令无关的令牌,在 Qwen2.5-VL-7B 上展示了显著的令牌减少和最小的准确性损失。ReFIT 提供了一个指令引导的框架,该框架适应相关区域的空间特征并进一步去除不必要的令牌,从而提高了 VQA 基准测试的准确性并降低了成本。 AI

影响 这些在视觉令牌减少方面的进展可以显著加快大型视觉语言模型的推理时间和降低计算成本,使其在实际应用中更易于访问和更高效。

排序理由 多篇研究论文介绍了 VLM 视觉令牌减少的新颖方法。

在 arXiv cs.CV 阅读 →

AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →

新方法优化大型视觉语言模型的视觉令牌化 · 跟踪到 4 个来源

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
多篇研究论文介绍了 VLM 视觉令牌减少的新颖方法。
Source corroboration
5 independent sources
Strong cross-source corroboration — multiple independent publishers covered this within the clustering window.
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
45 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.
Coverage growth since scoring
+1 source(s) since last score
New sources have picked up this story since our last re-score. Score will update on the next scoring pass.

完整方法见我们的编辑标准。

报道来源 [5]

  1. arXiv cs.AI TIER_1 English(EN) · Yihang Huang, Peisong Wen, Chenhao Zhou, Difu Feng, Qianqian Xu ·

    当1+1适得其反时:重新思考面向样本的视觉标记约简的相关性-多样性平衡

    arXiv:2610.16615v1 Announce Type: new Abstract: Visual token compression has emerged as a critical technique for reducing the inference cost of Vision-Language Models. Existing methods select tokens based on text relevance, visual diversity, or a fixed combination of both---treat…

  2. Hugging Face Daily Papers TIER_1 English(EN) ·

    MixTok:具有全局一维令牌和稀疏二维精炼的自适应视觉令牌化

    Efficient visual tokenization requires balancing representation compactness with the preservation of fine-grained image content. Conventional grid-based tokenizers allocate tokens uniformly across spatial locations despite non-uniform visual complexity, whereas compact 1D tokeniz…

  3. arXiv cs.CV TIER_1 English(EN) · Chenge Li, Yefei He, Xinyu Liu, Feng Chen, Hong Zhou, Bohan Zhuang ·

    MixTok:具有全局一维令牌和稀疏二维细化的自适应视觉令牌化

    arXiv:2610.13181v1 Announce Type: new Abstract: Efficient visual tokenization requires balancing representation compactness with the preservation of fine-grained image content. Conventional grid-based tokenizers allocate tokens uniformly across spatial locations despite non-unifo…

  4. arXiv cs.CV TIER_1 English(EN) · Yuna Lee, Kyoungho Min, Yulhwa Kim ·

    ERASE:通过自适应两阶段令牌剪枝消除冗余视觉令牌

    arXiv:2605.09982v2 Announce Type: replace Abstract: Recent advances in Vision-Language Models (VLMs) enable large language models (LLMs) to process high-resolution images, significantly improving real-world multimodal understanding. However, this capability introduces a large num…

  5. arXiv cs.CV TIER_1 English(EN) · Seyoung Jeong, Jong Pil Yun, Sang Jun Lee ·

    自适应视觉标记缩减以加速图像理解

    arXiv:2610.09252v1 Announce Type: new Abstract: Large Vision-Language Models achieve strong VQA performance, but processing high-resolution, information-rich images requires substantial computation, motivating visual token reduction. However, existing methods often prune individu…