Frozen Golden Set

A fixed, human-reviewed set of test cases used in machine learning evaluation to compare model performance in a stable and reproducible manner.