llm-internals

Why we included this project

This series explains how large language models work from the inside, starting with tokenization and Byte Pair Encoding and moving through the math behind Query, Key, and Value attention, causal masking, backpropagation, cross-entropy loss, and the Transformer architecture. Each post works through a concrete numeric example, and every topic has a video or written tutorial, so you can pick the format you prefer. It assumes you already have some machine learning background and want the mechanics spelled out rather than skimmed over. The author keeps adding new material, which makes the repository a useful index for self-study, whether you are learning the pieces in order or jumping straight to the part you need before training or optimizing your own models.