NinaCalvi/prefcollection-full-armo-rm

该数据集包含多个文本相关字段，如prompt（提示）、chosen（选择的回答）、rejected（拒绝的回答）等。此外，数据集还包含一个messages字段，用于存储对话内容及其角色信息。数据集还包含两个评分字段score_A和score_B，用于评估回答的质量。训练集包含72630个样本，总大小为855993261字节。

This dataset contains multiple text-related fields such as prompt, chosen (selected response), rejected (rejected response), etc. Additionally, the dataset includes a messages field for storing conversation content and role information. The dataset also contains two scoring fields, score_A and score_B, for evaluating the quality of responses. The training set consists of 72,630 samples with a total size of 855,993,261 bytes.