dl-model-advisor knowledge catalog
Bundle OKF 0.2 · 24 conceitos · Aznatkoiny/ai-toolkit-pro
Open source Repository Open in the app JSON
About
# dl-model-advisor knowledge catalog
The knowledge layer behind the dl-model-advisor plugin. Concepts carry OKF
provenance (`sources`), trust (`generated`/`verified`), and lifecycle
(`status`/`stale_after`). The advisor cites concept IDs; evidence tiers in
its plans derive from the `verified` entries here.
# Foundations (book canon — DLwP-2E, 2021)
* [Modality → architecture map](/foundations/modality-architecture-map.md) - routes data modality to architecture family
* [Text ratio rule](/foundations/text-ratio-rule.md) - the 1,500 samples/word threshold
* [Image data-size ladder](/foundations/image-data-size-ladder.md) - pretrained vs from-scratch by dataset size
* [Evaluation protocol by size](/foundations/evaluation-protocol-by-size.md) - holdout vs K-fold vs iterated K-fold
* [Universal workflow](/foundations/universal-workflow.md) - baseline → overfit → regularize
* [Pairing: binary classification](/foundations/pairing-binary-classification.md) - sigmoid + binary_crossentropy
*
Details
- Kind
- OKF bundles
- Topic
- Government & public data
- Publisher
- aznatkoiny
- Origin
- okf_github
- Category
- dados
- Version
- 0.2
- Open pull requests
- 4
- Last push
- 2026-09-07T00:39:11Z
- Repository state
- ativo
- Language
- Python
- Added
- 2026-09-09 12:03:57
- Updated
- 2026-09-09 12:03:57
- Origin id
Aznatkoiny/ai-toolkit-pro:knowledge/index.md