Dataset · Ajax102
Large-scale Chinese image-text corpus
Ajax102/Taisu
A massive collection of Chinese image-text pairs totaling approximately 7.9 TB for image-to-text tasks.
What's inside
The dataset contains between 100 million and 1 billion pairs of images and Chinese text descriptions. The total data volume is approximately 7.9 terabytes. Source files are packaged into tar archives of no more than 10 GB each.
What you can do with it
- Pre-train vision-language models (VLMs) for the Chinese language
- Train image captioning systems
- Tasks
- Image → text
- Languages
- Chinese
- Size
- 100M–1B rows
- Downloads this month
- 19.8K
- Likes
- 3
How to load it
from datasets import load_dataset
ds = load_dataset("Ajax102/Taisu")