FOUNDATIONS • 30 / 397
Start before LLMs. Understand where AI, ML and neural networks fit.
Gradient Descent
Gradient descent is the general idea of adjusting parameters in directions that reduce loss.
Think of it like
Think of Gradient Descent 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 Descent somewhere inside.
SRE lens
It belongs to training, not normal production inference.
Remember thisGradient descent is the general idea of adjusting parameters in directions that reduce loss.