---
task_categories:
  - image-text-to-text
---
# Caption3 Vision

Image-caption dataset from Caption3o-LongCap-v4.
- **image**: PIL image
- **text**: caption
- train: 100K rows, test: 6 rows