CValues-Comparison Chinese Large Model Value Evaluation Benchmark Dataset

With the rapid development of Chinese big models, their capabilities are constantly improving, and more and more people are beginning to worry about the risks they may bring. Therefore, the value assessment and value alignment around Chinese big models have received great attention. In order to promote research in this direction, Tongyi Lab has open-sourced the CValues-Comparison dataset.This is a dataset containing 145k value comparison samples, each sample contains a prompt, a positive response, and a negative response.

The data construction method can be found in the paper "CVALUES: Measuring the Values of Chinese Large Language Models from Safety to Responsibility"