PulseAugur
实时 10:21:25

HEAPr算法精确剪枝LLM专家,降低内存需求

研究人员开发了HEAPr,一种旨在减少混合专家(MoE)大型语言模型内存占用的新剪枝算法。与之前剪枝整个专家的旧方法不同,HEAPr将专家分解为更小的原子单元。这种方法利用原子专家输出的二阶信息,显著降低了计算复杂度,并实现了在精度损失最小的情况下进行更精确的压缩。在DeepSeek MoE和Qwen MoE等模型上的实验表明,HEAPr在高达25%的剪枝率下可以实现近乎无损的压缩,并将浮点运算次数(FLOPs)降低相似的幅度。 AI

影响 通过降低内存需求且不显著降低性能,从而能够更有效地部署大型MoE模型。

排序理由 这是一篇描述一种新的LLM剪枝算法的研究论文。[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 →

HEAPr算法精确剪枝LLM专家,降低内存需求

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
Tool
这是一篇描述一种新的LLM剪枝算法的研究论文。[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, model release
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
104 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准

报道来源 [1]

  1. arXiv cs.AI TIER_1 English(EN) · Ke Li, Zheng Yang, Zhongbin Zhou, Feng Xue, Zhonglin Jiang, Wenxiao Wang ·

    HEAPr: 基于 Hessian 的输出空间高效原子专家剪枝

    arXiv:2509.22299v3 Announce Type: replace-cross Abstract: Mixture-of-Experts (MoE) architectures in large language models (LLMs) deliver exceptional performance and reduced inference costs compared to dense LLMs. However, their large parameter counts result in prohibitive memory …