THE COLLECTION
Task examples
Follow an experiment, plan a procedure, repair scientific code, or operate a specialist application. Each example connects a concrete task to the evidence used to assess it.
Selected task examples. Browse documented inputs, expected outputs, and evaluation criteria. Runnable task packages and public downloads are forthcoming. Browse all 198 draft public task IDs →
No examples match these filters
Try a different capability, discipline, or search term.
Illustrations are generated for specific documented tasks; they are not task inputs or evaluated outputs. Tasks without a matching illustration show an input excerpt.