CUBE-MT (CUltural BEnchmark with Multimodal Transformations) is an extension to the CUltural BEnchmark for Text-to-Image models (CUBE). CUBE contains cultural artifacts across 16 countries (Austria, Denmark, Germany, Ireland, Mexico, Netherlands, Poland, Russia, Spain, Switzerland, Brazil, France, Italy, Japan, UK, USA) and 3 domains (cuisine, landmarks, art(cube, muse-IT)) extracted… See the full description on the dataset page:
https://huggingface.co/datasets/Bowl-of-porridge/CUBE-MT.
