AI Alphabet Soup — the dictionary

The open data behind AI Alphabet Soup: a clear, searchable dictionary of AI terms, written in plain language.

Every term swimming in AI's acronym broth — from skills and loop engineering to vLLM, SGLang, TensorRT-LLM, CUDA, and NIM — lives here as a small JSON file you can read, search, and contribute to.

What's in here

terms/            One JSON file per term (the dictionary)
categories.json   The canonical list of categories
schema.json       The JSON Schema every term must satisfy
CONTRIBUTING.md   How to add or fix a term

Each term looks like this:

{
  "slug": "rag",
  "term": "RAG",
  "acronym": "Retrieval-Augmented Generation",
  "category": "Patterns",
  "short": "Fetching relevant documents at query time and feeding them to the model as context.",
  "definition": "Retrieval-Augmented Generation grounds a model's answer in external knowledge ...",
  "example": null,
  "related": ["embedding", "vector-database", "context-window"]
}

Want to add a term?

Two ways, both welcome:

• Suggest one in 30 seconds — open a "Suggest a term" issue and fill in the form. No coding required.
• Add it yourself — read CONTRIBUTING.md, add a file under terms/, and open a pull request.

A GitHub Action checks every contribution against schema.json automatically, so you'll know right away if anything needs fixing.

License

The dictionary data is licensed under CC BY 4.0 — free to use and share with attribution.