Machines
How models learn
A course on how large language models are built and trained, written for curious people and working builders. About 2 hours of reading across 18 lessons, with 12 films to match.
Part 1
The basics
What a model is, how it reads text, and how it learns.
Part 2
Feeding the model
Data, scale and the memory a model needs.
Part 3
The machine room
Chips, parallelism and what goes wrong at scale.
Part 4
After pre-training
Turning a text predictor into something useful, and running it.
Part 5
End to end
Two long guides that follow one real model, Agora 9B, from raw text to a working assistant.
Films
Watch the whole story
How a model is trained
Data, scale, cost and the engineering behind a frontier run.
The open ecosystem
The strongest open models and what makes them tick.
The hardware
What the chips can and cannot do, and why.
How agents work
Models that use tools and act in the world.
Safety and alignment
Keeping capable models honest, understandable and under control.