item-response-theory
Everything on Ground Truth tagged “item-response-theory” — 2 items.
Study finds benchmark difficulty gaps can hide model progress News
A study of 25 benchmarks finds uneven difficulty coverage and reports that targeted, human-reviewed additions reduce measured gaps.
Item response theory: separating model ability from question difficulty Lesson
Item response theory estimates model ability and question difficulty together, helping explain what an AI benchmark measures and where it leaves gaps.