For software engineers learning about LLMs
Build a language model. Understand how it learns.
Start with word counts and build toward a small transformer trained on Shakespeare. Explore the examples in your browser, then run the Python project. Then adapt a pretrained model to answer questions with SQL. The ML fundamentals course explains the mathematics behind predictions and training.
Choose what you want to learn
Building Language Models from Scratch
Build a count-based predictor, learn how embeddings and attention use context, then train, save, and inspect a small Shakespeare transformer.
See the lessons → 5 lessons + optional Apple Silicon lab · Basic PythonFine-tuning a Model for SQL
Use WikiSQL to train a small adapter, inspect the training loop, and compare held-out answers with prompting. Learn why valid SQL can still be wrong.
See the lessons → 6 lessons · Basic algebraFundamentals of Machine Learning
Calculate a prediction, measure its error, and improve it. Learn loss, gradients, optimizers, batches, and probabilities with small worked examples.
See the lessons →Start without an account: read lessons, run prepared examples, use written hints, and keep explanation drafts on this device. Sign in for AI feedback, page chat on every lesson, and saved account progress. AI requests have usage limits. You can read any lesson without finishing the others.
Learning guides
- Six ways to learn AI — compare goals, prerequisites, practice and depth before choosing a course.
- A study companion to Karpathy’s Zero to Hero — plan the video sequence and check your understanding between coding sessions.
Optional references
The machine-learning slides and shared Python examples supplement the lessons. Use them for additional practice alongside these learning tracks.