Attention Is All You Need (Vaswani et al.)

Details

Photo

—

Name

—

Field

Machine learning

Year

2017

Core idea

Proposes the Transformer architecture, ditching recurrence and using attention mechanisms to handle sequences efficiently.

Why it stuck

The title sounds like a self-help book, but it quietly rewired half the tools I use at work and for side projects.

Read where

On a glowing phone screen, standing on a crowded 6 train, with bad service and too many acronyms.