WorkWander.tech
WorkWander.tech

BabyLM Challenge

BabyLM 2026: Sample-Efficient Pretraining Challenge

In conjunction with
Empirical Methods in Natural Language Processing 2026
October 24-29, 2026
HungExpo
Budapest, Hungary

NLP Machine Learning

A challenge on sample-efficient and cognitively plausible language model pretraining from small, developmentally realistic data. The constraint is the point: pretraining on roughly the amount of language a child hears, rather than on internet-scale corpora, tests whether current architectures are anywhere near as sample-efficient as human learners. Results consistently show they are not.

  For more information, visit the workshop website

More workshops at Empirical Methods in Natural Language Processing 2026