Documentation taxonomy
Every entry is organised in two levels: the top level follows the adopter’s journey; inside each domain, content is written as the standard document types from Diátaxis, ISO/IEC/IEEE 26514, and the Good Docs Project.
The four domains
Section titled “The four domains”| Domain | The question it answers | Typical document types |
|---|---|---|
| Assess | Should we adopt this? | Explanation, Model-Card intended-use & limitations, evaluation/benchmarks, license & openness, provenance, EU AI Act |
| Implement | How do we deploy it? | Quickstart, install/deploy how-tos, hardware/quantization reference, tutorial, fine-tuning |
| Use | What can it do and how? | Capabilities & prompt-template reference, task how-tos, concepts |
| Support | How do we keep it running? | Troubleshooting, release notes, security disclosure, community |
Why “Assess” is a first-class domain
Section titled “Why “Assess” is a first-class domain”Classic software-documentation frameworks (Diátaxis and the tutorial/how-to/reference/ explanation sites) assume you’ve already chosen the tool - they have no “should I adopt this?” category. But that decision is exactly what the AI/ML Model Card tradition documents (intended use, limitations, evaluation). For a library whose purpose is the adopt-or-not decision, Assess has to be first-class - a deliberate, well-precedented extension for models.
The full rationale and cross-framework grounding is in the repository.