ml/journey

An interactive, from-arithmetic-up path to understanding machine learning. Built on first principles. Designed to feel like figuring things out, not like reading a textbook.

Track 4

Learning machines

A single neuron, autograd, MLPs — the core of every modern model.

M10

A single neuron

The world's smallest learner.

4 lessons · ~33 min

M11

Autograd, from scratch

Build micrograd. The computational graph. Backprop as bookkeeping.

upcoming

M12

MLPs & nonlinearity

Stack neurons. Activation functions. Train on a real dataset.

upcoming

Track 5

Sequence models & language

Bigrams, embeddings, attention, transformers, GPT.

M13

Bigrams & makemore

The simplest language model. Counting vs learning.

upcoming

M14

Embeddings & tokens

Words become vectors. Tokenization.

upcoming

M15

Attention

Every token looks at every other token. Query / key / value.

upcoming

M16

Transformers & GPT

Stack attention + MLP blocks. nanoGPT-shaped finale.

upcoming