Every definition in the glossary sits in one subject area. Pick the one nearest your question and read outwards from there.
Each area is a self-contained reading path. The entries inside cross-link to the rest of the glossary, so you can start narrow and widen out.
The core ideas behind systems that learn patterns from data.
7 termsNeural networks with many layers, and the machinery that trains them.
5 termsHow machines read, generate and reason about human language.
7 termsModels that produce new text, images, audio or code.
4 termsMaking sense of images and video.
1 termDatasets, labelling, evaluation and the training loop itself.
1 termAlignment, bias, interpretability and the limits of these systems.
3 termsServing, scaling and running models in production.
4 termsThree entries from each area, so you can tell at a glance whether it is the one you want.
Topics cut across the areas — “Acronym” and “Beginner Friendly” each pull from several of them at once.