Zero to GPT-2

Build deep intuition for every concept behind GPT-2, from first principles. No hand-waving, no magic. Each chapter earns the next.

8 chapters·4 weeks·Vectors to transformers
Week 1Mathematical Foundations
01

Vectors & The Dot Product

The fundamental operation behind everything in ML.

02

Gradients & Learning

How machines learn by following the slope downhill.

Week 2Neural Networks
03

Neural Networks & Backprop

Stacking dot products to learn any function.

04

Embeddings: Words as Vectors

Turning language into numbers that capture meaning.

Week 3The Attention Revolution
05

Attention: The Breakthrough

The mechanism that changed everything about NLP.

06

Self-Attention: Q, K, V

The actual math behind attention, step by step.

Week 4Building GPT-2
07

The Transformer Block

Attention + normalization + feed-forward, one block.

08

GPT-2: The Full Picture

Architecture, training, generation. All of it.