Skip to content

Dataset · LooksJuicy

Chinese Ruozhiba questions with GPT-4 answers

LooksJuicy/ruozhiba

Curated Chinese questions with short-style GPT-4 answers inspired by COIG-CQIA.

What's inside

The dataset is based on COIG-CQIA: questions are taken from the Ruozhiba GitHub collection, and answers are generated by GPT-4. Answers were filtered to remove explicit refusals and made shorter than in COIG-CQIA. Data is in Chinese, sized between 1k and 10k examples.

What you can do with it

  • SFT for Chinese instruct models
  • Training concise answers to absurd questions
  • Evaluating refusal filtering in Q&A
Tasks
Text generation
Languages
Chinese
Size
1K–10K rows
Downloads this month
529
Likes
337

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("LooksJuicy/ruozhiba")