Sefika/

FewRel_Inverse_Relations

This dataset evaluates the robustness of relation extraction (RE) models with a focus on inverse relations and entity substitution. It combines examples from FewRel and TekGen to measure whether models correctly capture relation semantics independent of relation direction and entity identity. The dataset includes MLCommons Croissant metadata for improved interoperability with machine learning tools… See the full description on the dataset page: https://huggingface.co/datasets/Sefika/FewRel_Inverse_Relations.