English(EN) EFQ-Softmax: Exp-Free Quantization for Softmax

新的EFQ-Softmax方法优化Transformer的低比特量化

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-09 05:04

研究人员开发了EFQ-Softmax，一种用于Transformer模型低比特量化的新颖方法，它绕过了Softmax的传统指数计算。该方法直接将移位的注意力分数映射到E2M1操作数，在不牺牲模型质量的情况下优化效率。在Qwen3-8B和Qwen3-VL-8B-Instruct等模型上的评估显示性能指标有所提高，并且在A5向量单元上的内核级测试证明延迟显著降低。 AI

影响该方法可以实现更高效的大型语言模型在精度能力有限的硬件上的部署。

排序理由该集群描述了一种优化Transformer模型的新方法，该方法在一篇研究论文中有详细介绍。[lever_c_demoted from research: ic=1 ai=1.0]

在 Hugging Face Daily Papers 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: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群描述了一种优化Transformer模型的新方法，该方法在一篇研究论文中有详细介绍。[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: 5 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]

Hugging Face Daily Papers TIER_1 English(EN) · 2026-09-09 05:04

EFQ-Softmax：Exp-Free 量化 Softmax

Low-bit attention accelerates Transformer inference by moving the $QK^\top$ and $PV$ matrix multiplications to FP8 or FP4 matrix engines. However, the softmax path often evaluates shifted-score exponentials in higher precision, forms a temporary probability block, and quantizes i…

本文如何被排名

报道来源 [1]

EFQ-Softmax：Exp-Free 量化 Softmax

相关实体

相关话题