How Models Learn

Logit

A raw score a model gives one class before softmax; it can be any number, positive or negative.

It outputs one raw score per class, called a logit, and a logit can be any number, positive or negative.
From Softmax and cross-entropy, step “From scores to probabilities”

See it in action

Try logit hands-on in the lesson Softmax and cross-entropy, part of the The maths behind AI course.

Open the lesson