English(EN) The Ephemeral Cost of Switching Optimizers Across LLM Training Stages

研究发现：大型语言模型优化器切换的性能成本极小

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-19 04:00

一篇新的arXiv论文研究了在大型语言模型（LLM）训练过程中切换优化器的影响。研究人员发现，性能在很大程度上取决于预训练和后训练优化器的独立效果，而不是不匹配带来的显著惩罚。具体来说，即使模型使用AdamW进行预训练，在监督微调（SFT）期间增加Muon优化器的更新比例通常也能提高下游性能。虽然匹配预训练优化器可以减少SFT期间的遗忘，但并未带来更好的最终基准测试结果。 AI

影响表明在大型语言模型的后训练阶段选择优化器具有灵活性，而不会导致显著的性能下降。

排序理由关于大型语言模型训练方法的学术论文。[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 →

本文如何被排名

Signal score: 15 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 关于大型语言模型训练方法的学术论文。[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, 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: 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) · Xinyu Zhao, Davis Wertheimer, Ashutosh Sharma, Rameswar Panda, Paul Pu Liang · 2026-10-19 04:00

在LLM训练阶段切换优化器的短暂成本

arXiv:2610.22650v1 Announce Type: new Abstract: Recent work suggests that switching optimizers between pretraining and subsequent training stages incurs a performance cost, but systematic investigation of this phenomenon is lacking. We study the mismatch between AdamW and Muon op…

本文如何被排名

报道来源 [1]

在LLM训练阶段切换优化器的短暂成本

相关实体

相关话题