PulseAugur
实时 09:29:48

DAMP新技术大幅降低LLM内存使用并提升速度

研究人员开发了一种名为DAMP(Decay-Aware Mixed-Precision Recurrent-State Quantization)的新型量化技术,以减少使用循环状态的大型语言模型的内存占用并提高其速度。传统模型将这些状态存储为FP32,消耗大量GPU内存并增加延迟。DAMP识别这些状态中的高风险通道,并以更高精度存储它们,同时将其余部分量化为INT8,在接近FP32的精度下,存储量减少了69.1%,循环状态更新速度提升高达2.01倍。该方法在Qwen3.6-35B和Kimi-Linear-48B等模型上进行了评估,证明了其在各种推理和代码生成任务中的有效性。 AI

影响 降低了具有循环状态的LLM的内存和延迟,可能在现有硬件上实现更大的模型或更快的推理。

排序理由 详细介绍LLM量化新技术的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

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

DAMP新技术大幅降低LLM内存使用并提升速度

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LLM量化新技术的学术论文。[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]

  1. arXiv cs.LG TIER_1 English(EN) · Tao Zhang, Jianchao Tan, Pingwei Sun, Yanqi Yu, Zixu Jiang, Yuchen Xie, Xunliang Cai, Ziqian Zeng ·

    DAMP:衰减感知混合精度循环状态量化

    arXiv:2608.27513v1 Announce Type: new Abstract: Softmax attention stores key and value vectors for every preceding token, causing inference memory to grow with sequence length. Recent language models incorporating Gated DeltaNet (GDN) or Kimi Delta Attention (KDA) reduce this cos…