mlfoundations-dev/science_verification_scale_up_random

该数据集包含了问题、推理过程、解决方案等信息的文本数据集。数据集中的每个样本都包含了用户面临的问题、用户的推理过程、DeepSeek工具提供的解决方案、原始行索引、最终推理轨迹、大多数用户的响应、验证后的最终推理轨迹以及相关对话。数据集分为训练集，共有48000个示例，大小为923,792,057字节。

This dataset is a text dataset containing information such as problems, reasoning processes, and solutions. Each sample in the dataset includes the problem faced by the user, the users reasoning process, the solution provided by the DeepSeek tool, the original row index, the final reasoning trace, the majority of user responses, the verified final reasoning trace, and related conversations. The dataset is divided into a training set with a total of 48,000 examples, totaling 923,792,057 bytes in size.