arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.02873cs.CL
收藏

LaTeX编译:在大语言模型时代面临的挑战

.tmu: A Low-Entropy Tree-Structured Representation for LLM-Assisted Scientific Writing

Southern University of Science and Technology(南方科技大学);Alibaba(阿里巴巴);Tsinghua University(清华大学);Rutgers University(罗格斯大学);Liii Network(Liii网络)

Tianyou Liu, Ziqiang Li, Xurui Liu, Yu Wu, Yansong Li

AI总结:

本文提出Mogan STEM作为替代TeX的结构编辑器,通过高效的数据结构和按需插件加载,在编译效率、生成语义和工具生态系统等方面优于TeX,并展示了其在LLM微调中的优势。

摘要:

随着大语言模型(LLMs)越来越多地协助科学写作,TeX的局限性和显著的令牌成本变得越来越明显。本文分析了TeX在编译和用户体验设计中的基本缺陷,以说明其在大语言模型时代在编译效率、生成语义、错误定位和工具生态系统方面的局限性。作为替代方案,Mogan STEM,一个WYSIWYG结构编辑器,被引入。Mogan通过其高效的的数据结构、快速的渲染和按需插件加载,在上述方面优于TeX。进行了广泛的实验,以验证在LLM任务中的编译/渲染时间和性能的优势。此外,我们表明,由于Mogan的较低信息熵,使用Mogan的.doc格式(Mogan的文档格式)对LLM进行微调比TeX更高效。因此,我们呼吁在使用.doc格式进行LLM训练时进行更大规模的实验。

英文摘要:

As large language models (LLMs) increasingly assist scientific writing, the limitations and token costs of generating TeX become increasingly visible. This paper analyzes TeX's architectural mismatch with LLM workflows, stemming from its lack of an explicit structural representation, to illustrate its limitations on generated semantics and error localization. As an alternative, we introduce .tmu, a low-entropy tree-structured representation. With its efficient data structure and clear contextual boundaries, .tmu outperforms .tex in the above aspects. Experiments across four LLMs provide evidence for this claim in most evaluated settings. Furthermore, we show that due to its lower information entropy, fine-tuning LLMs on .tmu achieves approximately 43% lower final training loss than on .tex. Our work provides a more scalable and LLM-friendly data representation for LLM-assisted scientific writing.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑