Skip to content

Dataset · IDEA-CCNL

Chinese Subset of the LAION-2B Multimodal Dataset

IDEA-CCNL/laion2B-multi-chinese-subset

About 143 million Chinese image-text pairs from LAION-2B for multimodal tasks.

What's inside

A Chinese-language subset of the large multilingual LAION-2B multimodal dataset, prepared by the Fengshenbang-LM project. It contains about 143 million text and image URL pairs (images are not included, only text metadata totaling about 19 GB).

What you can do with it

  • Training Chinese multimodal models (e.g., CLIP counterparts)
  • Pretraining text-to-image generation models in Chinese
  • Feature extraction and cross-modal retrieval for Chinese
Tasks
Embeddings
Languages
Chinese
Size
10M–100M rows
Downloads this month
308
Likes
42

How to load it

from datasets import load_dataset

ds = load_dataset("IDEA-CCNL/laion2B-multi-chinese-subset")