Multi-CrowS-Pairs is a multilingual extension of the CrowS-Pairs dataset, specifically designed for measuring stereotypical biases in masked language models across multiple Indian languages. This dataset includes translations in Hindi and Bengali, alongside the original English version, with gender-specific pronoun annotations to ensure accurate bias measurement in languages with different grammatical structures.… See the full description on the dataset page: https://huggingface.co/datasets/Debk/Multi-CrowS-Pairs.
