How does a
neural network learn?
Build intuition for predictions, errors and learning. Then train a small classifier and test what it really learned.
Basic algebra and graph reading. No coding required for the guided route.
Your route through the ideas
Open any unit. Earlier lessons are recommendations, not locks.
01How does a network learn?12 lessons
Start with a single neuron. Discover how small numerical steps become learning.
02What has the network learned?13 lessons
Look inside learned features and find the shortcuts that can fool a model.
03How does AI process language?13 lessons
Follow a word from a numerical representation to a next-token prediction.
04Why does AI need computation?8 lessons
Understand the tradeoffs between model size, data, and computation.
05Data audits and training diagnosis10 lessons
Repair an evaluation pipeline and diagnose a training failure, with separate project assessments.
06Simple models and useful baselines10 lessons
Compare neighbors, fitted trees, numerical regression and class probabilities with simple references before adding complexity.
07Generative models beyond text8 lessons
Learn reconstruction, latent representations and the limits of sampling from small trained models.