StereoSet

Dataset Card for StereoSet Dataset Summary StereoSet is a dataset that measures stereotype bias in language models. StereoSet consists of 17,000 sentences that measures model preferences across gender, race, religion, and profession. Supported Tasks and Leaderboards multiple-choice question answering Languages English

Papers using StereoSet (7)

A Comprehensive Study of Implicit and Explicit Biases in Large Language Models2025No Free Lunch in Language Model Bias Mitigation? Targeted Bias Reduction Can Exacerbate Unmitigated LLM Biases2025Addressing Stereotypes in Large Language Models: A Critical Examination and Mitigation2025Open-DeBias: Toward Mitigating Open-Set Bias in Language Models2025Rethinking Prompt-based Debiasing in Large Language Models2025Shifting Perspectives: Steering Vectors for Robust Bias Mitigation in LLMs2025Mitigating Social Bias in Large Language Models: A Multi-Objective Approach within a Multi-Agent Framework2024 · 1 cites