English(EN) Beyond Rephrasing: Book-Level Organization Improves Synthetic Textbook Data for Mid-Training

书籍级组织提升了用于LLM训练的合成教科书数据

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-01 04:00

研究人员开发了一种新的合成教科书数据创建方法，显著改善了语言模型的训练。该方法将相关内容组织成连贯的书籍级文档，这是以前被忽视的因素，而只关注局部改写。该流程生成了超过68.6万本教科书，在下游任务上的平均性能提高了1.09。实验表明，书籍级组织，而不仅仅是内容或长度，是这种性能提升的关键，其表现优于随机拼接和独立章节改写。 AI

影响这项研究提出了改进LLM训练数据的新途径，有望以更低的计算成本带来更强大的模型。

排序理由该集群包含一篇学术论文，详细介绍了用于语言模型训练的合成数据生成新方法。[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: 25 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇学术论文，详细介绍了用于语言模型训练的合成数据生成新方法。[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) · Jiawen Tao, Miao Peng, Yaoming Li, Xiaokun Yuan, Mengzhou Wu, Wenhan Yu, Guoan Wang, Nuo Chen, Tong Yang, Maxm Pan · 2026-10-01 04:00

超越改写：书籍级组织改进了用于中期训练的合成教科书数据

arXiv:2607.28109v2 Announce Type: replace Abstract: Synthetic textbook data has improved language model pre-training, but prior work largely treats the benefit as a property of generated content or local rewriting style. We study a different factor: whether related content is org…

本文如何被排名

报道来源 [1]

超越改写：书籍级组织改进了用于中期训练的合成教科书数据

相关实体

相关话题