English

Training a T5 Using Lab-sized Resources

Computation and Language 2022-08-26 v1

Abstract

Training large neural language models on large datasets is resource- and time-intensive. These requirements create a barrier to entry, where those with fewer resources cannot build competitive models. This paper presents various techniques for making it possible to (a) train a large language model using resources that a modest research lab might have, and (b) train it in a reasonable amount of time. We provide concrete recommendations for practitioners, which we illustrate with a case study: a T5 model for Danish, the first for this language.

Keywords

Cite

@article{arxiv.2208.12097,
  title  = {Training a T5 Using Lab-sized Resources},
  author = {Manuel R. Ciosici and Leon Derczynski},
  journal= {arXiv preprint arXiv:2208.12097},
  year   = {2022}
}
R2 v1 2026-06-25T01:58:31.375Z