Learning by building low-level AI tools

Learning path

Work through the tools in order — each page has a playground above and a Learn section below with docs and diagrams.

1. 01

   Tokenizer Explorer

   How text becomes token IDs — the first step in every LLM.

2. 02

   Matrix Playground

   The linear algebra underneath neural networks.

3. 03

   Embedding Explorer

   How meaning becomes vectors you can compare and search.

4. 04

   Attention Visualizer

   How tokens decide which other tokens to focus on.

5. 05

   Transformer Builder

   How attention and feed-forward layers stack into a model.

6. 06

   GGUF Inspector

   What's inside a real model file on disk.

Tools

• Tokenizer ExplorerLive

  Paste text, inspect tokens, BPE merges, and vocabulary lookups.

• Matrix PlaygroundLive

  Visualize matrix operations that power neural networks.

• Embedding ExplorerLive

  See how text becomes vectors and how similarity works.

• Attention VisualizerLive

  Watch attention weights connect tokens in a sequence.

• Transformer BuilderLive

  Assemble transformer blocks and trace the data flow.

• GGUF InspectorLive

  Peek inside quantized model files for local inference.