Training compute-optimal transformer encoder models.

Megi Dervishi, Alexandre Allauzen, Gabriel Synnaeve, Yann LeCun

Browse the full EMNLP paper archive.