Skip to content

benchmark

FalseQA

Question-answering dataset built from questions that assume a false fact, used to test whether a model rejects the premise instead of answering.

Current clusters