attention

어텐션

A core computation in AI models that calculates how relevant each word is to every other word in a passage, letting the model focus on the important parts. It underpins today's dominant transformer models, but its computational cost grows sharply as text gets longer.

Briefings mentioning this
2026-08-11