English(EN) DualQuant: Dual-scale Calibration-free Weight-and-Activation Quantization for LLMs

DUALQUANT框架优化大语言模型，实现无校准低比特推理

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

研究人员开发了DUALQUANT，一个新颖的训练后量化框架，旨在优化大语言模型（LLMs）在加速器上进行低比特推理。该方法通过最小化权重重建均方误差（MSE）来实现具有竞争力的准确性，而无需校准数据，同时还能提供有效的逐通道激活重缩放。DUALQUANT支持MX和NVFP4等多种格式，并在Llama和Qwen等模型上展示了显著的内存减少和吞吐量加速。 AI

影响使得在具有原生低比特支持的硬件上更高效地部署大语言模型成为可能，从而可能降低推理成本并提高速度。

排序理由该集群包含一篇详细介绍大语言模型量化新方法的论文。[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: 12 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍大语言模型量化新方法的论文。[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) · Mahsa Salmani, Sayeh Sharify, Hesham Mostafa · 2026-10-12 04:00

DualQuant：LLM 的双尺度无校准权重和激活量化

arXiv:2610.13767v1 Announce Type: cross Abstract: Low-bit inference on accelerators with native 4-bit support requires quantizing both weights and activations to realize the full throughput gains these formats provide. While post-training weight quantization has been extensively …

本文如何被排名

报道来源 [1]

DualQuant：LLM 的双尺度无校准权重和激活量化

相关实体

相关话题