• Math3ma celebrates 10‑year anniversary, reflecting on growth from niche blog to influential math resource. • Author began as study tool, now shares graduate‑level insights with global audience. • New preprint explores category theory framework for language, inspired by LLMs’ text corpora. • Paper defines magnitude of categories of texts enriched by language models, linking mathematics and AI. • Collaboration with Juan Pablo Vigneaux builds on earlier 2021 work with Terilla and Vlassopoulos. • Blog highlights how LLMs provide rich structure for mathematical modeling of language.
Article Summaries:
- Math3ma, a mathematics blog founded in 2015, celebrates its tenth anniversary, noting its growth from a personal study tool to a widely visited site. The blog’s author, now collaborating with Juan Pablo Vigneaux, has released a new preprint on arXiv titled The Magnitude of Categories of Texts Enriched by Language Models. The paper builds on a 2021 work that applied category theory to language, modeling text as a category where morphisms represent substring containment. It extends this framework by incorporating conditional probabilities derived from large language models, aiming to capture statistical relationships between strings within corpora.
Sources: