ACIDE/user-vlm-elderly-bench

这是一个针对特定领域评估的任务基准，包含144个图像、问题和答案的三元组。这些数据聚焦于老年人现实生活中与科技相关的问题，问题由巴黎公共援助医院提供，图像来自FairFace，确保了种族和性别的多样性。参考答案是通过GPT-4o生成的，提供了高质量的响应。

This is a task benchmark for domain-specific evaluation, comprising 144 triplets of images, questions, and answers. The data focuses on real-world technology-related queries from elderly individuals, with questions provided by Assistance publique Hôpitaux de Paris (APHP) and images from FairFace ensuring diversity in ethnicity and gender. Reference answers are generated using GPT-4o with detailed instructions for high-quality responses.