metadata
task_categories:
- image-to-text
- text-to-image
language:
- ar
- en
- ru
- it
- es
- ko
- pl
- tr
- zh
- de
- fr
- ja
size_categories:
- 1K<n<10K
Dataset Summary
The expanded XTD-11 dataset, now including Arabic, enhances the original XTD collection. This dataset introduces a 1,000-image multi-lingual MSCOCO2014 caption to test multimodel in zeroshot image or text retrieval in 11 Languages.