notesum.ai

Published at December 6

Findings of the Second BabyLM Challenge: Sample-Efficient Pretraining on Developmentally Plausible Corpora

cs.CL

Released Date: December 6, 2024

Authors: Michael Y. Hu1, Aaron Mueller2, Candace Ross3, Adina Williams3, Tal Linzen1, Chengxu Zhuang4, Ryan Cotterell5, Leshem Choshen6, Alex Warstadt7, Ethan Gotlieb Wilcox5

Aff.: 1New York University; 2Northeastern University; 3Meta AI; 4MIT; 5Georgetown University; 6IBM Research; 7ETH Zürich

Arxiv: http://arxiv.org/pdf/2412.05149v1