NiuTrans/GRAM-fine-tuning-65k

这是一个用于微调GRAM模型的数据集。数据集中的每一项包含了评价两个AI助手回答质量的指示，指示中包含了用户问题以及两个助手的回答。用户需要根据回答的有用性、相关性、准确性、深度、创造性和详细程度来评价哪个助手的回答更好，并输出A或B作为评价结果。数据集从Skywork-Reward-Preference-80K-v0.2中过滤得到，移除了过长和包含乱码的数据。

This is a dataset for fine-tuning the GRAM model. Each item in the dataset includes instructions for evaluating the quality of answers provided by two AI assistants to a users question, along with the assistants responses. Users are required to evaluate which assistants answer is better based on the helpfulness, relevance, accuracy, depth, creativity, and detail of the responses, and output A or B as the evaluation result. The dataset is filtered from Skywork-Reward-Preference-80K-v0.2, removing data that is too long and those containing garbled characters.