FOUNDATIONS • 29 / 397
Start before LLMs. Understand where AI, ML and neural networks fit.

Gradient

A gradient indicates how a small change to a parameter would change the loss.

Think of it like

Think of Gradient as one layer in a very large toolbox: you only need to understand what job that layer performs.

Real life

A phone spam filter, recommendation engine, fraud detector or voice assistant may use Gradient somewhere inside.

SRE lens

Training systems compute gradients; inference systems usually do not.

Remember thisA gradient indicates how a small change to a parameter would change the loss.
AIForSREJump to a concept
Search is optional. The main journey is simply ↓ Next.