AlignmentResearch/WildChat

该数据集包含了文本内容及其相关特征，用于训练文本分类模型，以区分文本内容是否为良性或有害。数据集分为default、neg和pos三个配置，每个配置都包含clf_label（分类标签）、instructions（指令）、content（内容）、completion（完成）、answer_prompt（答案提示）、proxy_clf_label（代理分类标签）、gen_target（生成目标）和proxy_gen_target（代理生成目标）等字段。clf_label字段标记为良性或有害。数据集包含训练集，但验证集大小为0。default和neg配置的训练集包含相同数量的示例，而pos配置的训练集为空。

This dataset includes text content and its related features for training text classification models to distinguish whether the content is Benign or Harmful. The dataset is divided into three configurations: default, neg, and pos, each containing fields such as clf_label (classification label), instructions, content, completion, answer_prompt, proxy_clf_label, gen_target, and proxy_gen_target. The clf_label field is marked as Benign or Harmful. The dataset includes a training set, but the validation set size is 0. The training sets of the default and neg configurations contain the same number of examples, while the training set of the pos configuration is empty.