Atlas

Large Language Models

Tokenization

Turning text into the integers a model can eat.

You can jump straight in, but this star assumes Embeddings. Not recommended as a first stop.

Four layers of depth

Each layer ends with a quiz. Finish layer 4 and you own this concept.

  1. L1IntuitionWhy a model can't read text at all — it reads integers — and how those integers get chosen.6m
    40 XP
  2. L2MechanicsHow BPE is actually trained and applied, byte-level fallback, special tokens, and vocabulary design trade-offs.9m
    70 XP
  3. L3CodeBuild a working byte-level BPE trainer and encoder from scratch, then compare it to a real tokenizer.14m
    110 XP
  4. L4FoundationsThe formal objective BPE greedily approximates, entropy and compression bounds, and the hardware cost of vocabulary size.12m
    180 XP

Where this leads

33 stars in the atlas.