penfever/dpo-qwen2572b-llama3170b-jdg-Llama3-Harmlessness

该数据集是针对对话系统的，包含问题、选择的回答、被拒绝的回答以及每个回答的评分信息，评分包括熵加权的峰度、平均值和方差。数据集分为训练集，共有约303万280个示例，数据大小为约1.86亿字节。

This dataset is for dialogue systems, including questions, chosen responses, rejected responses, and scores for each response, which include entropy-weighted kurtosis, mean, and variance. The dataset is split into a training set with approximately 3,032,080 examples, totaling about 1.86 billion bytes in size.