Runnable Python guides

AI Concepts in Python

Learn how neural networks, machine learning, Transformers and classical AI work with short runnable Python lessons that use no AI libraries.

14 runnable examples

Build an artificial neuron in plain Python: a weighted sum of inputs plus a bias, passed through sigmoid or ReLU.

Pass inputs through a small neural network layer by layer, then measure the error with squared error and cross-entropy.

Train a neuron with gradient descent: use the chain rule to find how each weight affects the loss, then step downhill.

Compare plain gradient descent, momentum and Adam by minimising the same function step by step.

Scale features with min-max normalization and z-score standardization, and multiply matrices by hand.

Fit a straight line with least squares, then turn a score into a probability with logistic regression.

Turn scores into class probabilities with softmax, score them with cross-entropy and classify points with k-nearest neighbours.

Measure how similar two vectors are with cosine similarity and group unlabelled data with k-means.

Turn text into numbers the way language models do: build a vocabulary, one-hot encode words and learn subword tokens with BPE.

Compute scaled dot-product attention by hand with queries, keys and values, then add a causal mask like GPT.

See how a language model picks the next token: layer normalization, temperature and top-k sampling in plain Python.

Find the shortest path on a grid with A* search and pick the best move in a game with minimax.

Predict future states with a Markov chain and balance exploring and exploiting with an epsilon-greedy bandit.

Teach an agent to reach a goal: compute state values with value iteration, then learn them by trial and error with Q-learning.