Deep Learning Basics
The core mechanics every practitioner should own — from a single neuron to training a network that generalizes.
Who this is for
Anyone who can code but treats neural-network training as a black box. This path makes the training loop concrete.
How to use this path
Do the math for a tiny network by hand once — two inputs, one hidden layer. Everything after that (optimizers, regularization, initialization) is a variation on the same forward/backward rhythm.
What you'll be able to do
Explain why a model isn't learning, read a loss curve, and reach for the right fix — more data, regularization, a different learning rate, or a better architecture.