Dataset · BAAI
CCI 3.0 HQ Chinese high-quality web corpus
BAAI/CCI3-HQ
A large cleaned Chinese text corpus from trusted internet sources for pre-training and fine-tuning language models.
What's inside
The dataset contains Chinese texts collected from trusted internet sources. This is the HQ part of CCI 3.0: compared to CCI, sources are expanded and filtering is stricter. Size ranges from 10 to 100 million entries; the format is not specified in the card.
What you can do with it
- Pre-training a Chinese language model
- Fine-tuning or adapting LLMs in Chinese
- Evaluating the filtering quality of the Chinese corpus
- Tasks
- Text generation
- Languages
- Chinese
- Size
- 10M–100M rows
- Downloads this month
- 4.5K
- Likes
- 70
- Access
- on request on Hugging Face
Models trained on it
- nicoboss/PLM-1.8B-Instruct-non-MLA-i1-GGUF 297
- nicoboss/PLM-1.8B-Instruct-non-MLA-GGUF 144
- PLM-Team/PLM-1.8B-InstructText generation 122
- TurboPascal/ChineseModernBert 69
- PLM-Team/PLM-1.8B-BaseText generation 37
- BAAI/Aquila-135MText generation 4
- BAAI/Aquila-135M-InstructText generation 4
- BAAI/CCI3-HQ-Intermediate-Checkpoints 0
How to load it
from datasets import load_dataset
ds = load_dataset("BAAI/CCI3-HQ")