LLM from scratch
Build the parts of a small language model in Rust, from tokenization to training and generation. Chapters 0–39.
Start LLM from scratchFrom text to a working model
Build an LLM from first principles
Follow small, executable examples until tensors, attention, training, and generation fit together as one Rust program.
Choose a course
Practical LLM in Rust
Extend the completed model with corpus preparation, efficient execution, and measured training and inference. Start with the course structure in Chapter 0.
Requires the basics from LLM from scratch or equivalent Rust and LLM experience.
Available in English
Start Practical LLM in RustLearn by running code
Every concept begins with a tiny Rust example and deterministic output.
Connect code to the formula
Symbols are introduced only when they explain the behavior you can observe.
See what came before
Each chapter contrasts the modern component with the historical approach it replaced.