Module 01

How LLMs Work

00Welcome: You Are Going to Build AI Systems01What Actually Is AI?02What Is a Model?03How Does a Model Learn?04Training, Validation, Testing, Inference05Neural Networks Without the Scary Math06Tokens: LLMs Don't Read Text Like You Do07Context Windows08Embeddings: Meaning as Numbers09Why Transformers?10Attention11Query, Key, Value12Self-Attention13Inside a Transformer Block14Encoder, Decoder, and Decoder-Only Models15What Is an LLM?16The Flagship Next-Token Simulation17Logits, Probabilities, Softmax18Sampling19Why Can an LLM Follow Instructions?

Step 01 / 03

What is inside many modern AI models?

Module map

You know a model is a learned function with parameters. What structure lets many modern systems learn complex patterns from messy inputs like language and images?

What structure is inside many modern AI models?