Concept
Early-exit
Early-exit is an inference technique where a model stops computing as soon as it is confident enough, rather than running every layer or iteration for every input. Think of it like short-circuit evaluation in a boolean expression, except the "condition" is a learned…
The rest of “Early-exit” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→