Specimen, sample rows
The obligation register
This is the layout, not the full kit. The paid kit carries every unit, the evidence list, the dated plan, the generative AI working papers and the CSV and XLSX versions.
| Unit | Title | Evidence artefact | Status | Owner | Due |
|---|---|---|---|---|---|
6.2 | Task | The problem, input format, output format and success measure. | Open | ||
6.5.3.2.9 | Transformer architecture and algorithm | Encoder-only, decoder-only or encoder-decoder, with the task it serves. | Open | ||
6.5.3.2.10 | Generative model | Why each system is or is not generative under the learned-distribution test. | Open | ||
7.5 | Self-supervised machine learning | The policy on using generated content in training and fine-tuning. | Open | ||
8.2 | Data acquisition | Source, method, date range and the population the data represents. | Open | ||
1 | Scope | A signed note that records whether the framework is adopted, and for which systems. | Open | ||
6.5.3 | Categories of machine learning algorithms | The family, the reason for the choice and the alternatives rejected. | Open | ||
6.5.3.2.10 | Generative model | Model card with the learned-distribution test and the family | In progress | AI governance lead | 2026-11-30 |
What the columns mean
Unit and title identify the framework unit and the page. Evidence artefact is what a reviewer samples. Owner, due date, status and evidence pointer are the fields your team fills in. The description, the question and the common gaps sit beside them in the full register.
Specimen, not a certificate. Not legal advice; does not certify.