AriseKit

Benchmark Tasks

A breakdown of the 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.

Explore

Tasks
All 117 benchmark-building tasks, grouped by phase and filterable by flag (A / T / S / N). Includes flag distribution and per-phase density views.
Open Tasks →
Flow
Phase-by-phase flow diagrams showing the loops where a downstream answer forces revising an earlier decision. Task numbers match the Tasks page.
Open Flow →