This dataset is Multi-round conversations, including formal and informal styles, covering a total of 19 domains, covering scenarios such as work, life, and school.
Original questions created by experts and scholars, covering mathematics, physics, chemistry, biology, and informatics. They have passed the initial screening. The reasoning is relatively rigorous and the provided steps can be verified. They focus on the core content from senior undergraduate to graduate level, and can effectively test the upper limit of the model's reasoning. The questions are mainly in Chinese, with a few purely English ones, and standard LaTeX symbols are embedded.
