"A Mixture of h - 1 Heads is Better than h Heads."

Hao Peng et al. (2020)

Details and statistics

DOI: 10.18653/V1/2020.ACL-MAIN.587

access: open

type: Conference or Workshop Paper

metadata version: 2024-02-21

a service of  Schloss Dagstuhl - Leibniz Center for Informatics