Dataset · Iess
Modern Chinese poetry and image-based pairs
Iess/chinese_modern_poetry
Corpus of 100K to 1M poems by modern Chinese and foreign poets with a selection for image-based generation.
What's inside
The dataset includes original poems by modern Chinese authors and translations of foreign poets into Chinese, gathered from open web sources (GitHub, Baidu, Zhihu, poetry websites). Metadata indicates a volume of 100,000 to 1,000,000 records. Data is presented in jsonl and json files, including a special selection for poem generation based on 2-5 key images (up to 256 tokens long), which was used to test ChatGLM and LLaMA-7B training.
What you can do with it
- SFT or LoRA fine-tuning for Chinese poetry generation
- Training models to compose poems based on a given set of key images
- Languages
- Chinese
- Size
- 100K–1M rows
- Downloads this month
- 780
- Likes
- 31
How to load it
from datasets import load_dataset
ds = load_dataset("Iess/chinese_modern_poetry")