kino is a course in AI, machine learning and deep learning. Learn how models become less wrong, from a single dot product to a transformer, by reading the idea, passing the test and writing the code yourself.
Vectors, gradients and probability: the three tools every model is built from.
Learning a mapping from labelled examples: regression, classification, neighbours, trees and ensembles.
Overfitting, regularisation, honest evaluation, and learning from data without labels.
From a single neuron to backpropagation, and the tricks that make deep networks train.
Convolutions for images, recurrence for sequences, and attention, the idea behind transformers.
How large language models and image generators work: tokens, next-token prediction, scaling, alignment and diffusion.
Short articles with interactive figures you can drag, and code cells you can edit and run on the spot.
Six questions on the ideas. Pass at 80%, retake as often as you like, and read why each answer is right.
Write a function in Python. Hidden checks run against it in your browser and tell you exactly what's off.
Every module ends with a guided build on real data: a line fit, a classifier, a neural network, a language model.
def softmax(z): z = z - z.max(axis=1, keepdims=True) e = np.exp(z) return e / e.sum(axis=1, keepdims=True)
Python runs inside the page through WebAssembly, so your code never leaves your computer and a runaway loop can be stopped with one click.
The Kino Certificate in AI/ML is issued when every theory test, practical and project passes. It carries the first and last name on your account, the date you finished, and a certificate number.
See it full size →Andrew Ng's machine learning course: the derivations behind regression, classification and generalisation.
Convolutional networks and backpropagation, taught by building them in numpy.
The open textbook used at hundreds of universities, from linear algebra to transformers.
Andrej Karpathy's series that builds autograd and a GPT from an empty file.
Lesson 1.1 takes nine minutes and ends with your first lines of numpy.
Start the course →