The full list of 117 tasks involved in building a clinical LLM benchmark, tagged by whether AriseKit can automate the work, serve as a tool for a human decision, provide a scaffold of templates, or sits outside our scope.
| Phase | A | T | S | N | Total | Mix |
|---|---|---|---|---|---|---|
| 1. Conception | 9 | 1 | 10 | |||
| 2. Scope | 12 | 12 | ||||
| 3. Data | 3 | 7 | 2 | 6 | 18 | |
| 4. Items | 10 | 6 | 1 | 17 | ||
| 5. Rubric | 5 | 15 | 20 | |||
| 6. Panel | 2 | 6 | 3 | 2 | 13 | |
| 7. Eval Run | 6 | 4 | 10 | |||
| 8. Release | 2 | 2 | 1 | 5 | ||
| 9. Governance | 5 | 1 | 6 | |||
| 10. Maintenance | 2 | 3 | 1 | 6 |