This dataset contains culturally diverse images and prompts for evaluating cultural competence in Vision-Language Models (VLMs), as presented in the paper Toward Socially Aware Vision-Language Models: Evaluating Cultural Competence Through Multimodal Story Generation.
Code:
https://github.com/ArkaMukherjee0/mmCulturalConcepts: 35 unique concepts (e.g., honesty, empathy, cooperation)
Cultural… See the full description on the dataset page:
https://huggingface.co/datasets/ArkaMukherjee/mmCultural.
