English(EN) Temperon: Full-Time SAM Quality at a Third Less Wall-Clock

Temperon训练方法以更低成本实现SAM质量

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-17 04:00

研究人员推出了一种新颖的训练方法Temperon，旨在以显著降低的计算成本实现Sharpness-Aware Minimization (SAM) 的质量。Temperon采用两阶段方法：初始探索阶段使用标准SGD，然后在训练后期切换到SAM包装的Muon精炼器。该方法在CIFAR-10/100和Tiny ImageNet等各种数据集上已证明具有与全时SAM相当的准确性，并能更快地达到目标准确率。该方法在预训练GPT-2和针对GLUE任务进行微调方面也显示出有效性，节省了大量实际运行时间。 AI

影响该方法可以显著减少训练大型模型所需的计算资源，使先进技术更加易于获取。

排序理由该集群包含一篇详细介绍机器学习模型新训练方法的论文。[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 →

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; 该集群包含一篇详细介绍机器学习模型新训练方法的论文。[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.LG TIER_1 English(EN) · Stamatis Mastromichalakis · 2026-09-17 04:00

Temperon：以三分之一的墙上时钟时间实现全职SAM质量

arXiv:2609.17575v1 Announce Type: new Abstract: Sharpness-aware minimization (SAM) doubles the cost of every training step, yet its benefit concentrates where training ends. We study where an expensive training mode should be spent and propose Temperon: a plain-SGD explorer for t…

本文如何被排名

报道来源 [1]

Temperon：以三分之一的墙上时钟时间实现全职SAM质量

相关实体

相关话题