Neural Machine Translation by Jointly Learning to Align and Translate
At a glance
Attention helps neural translation models focus on relevant words as each word is generated.
In one paragraph
Attention lets a translation model focus on relevant source words as it generates each translated word. This avoids squeezing an entire sentence into one fixed summary. The approach improved English-to-French translation, particularly for longer sentences that challenged earlier models.
Authors 3
Organizations
Credited for this work at publication.