selfcorrexp2/w2r125k_r2r90k_r125k

这是一个包含对话和解决方案等信息的训练数据集，适用于文本生成和对话系统等领域。数据集包括了索引、提示、是否为第一轮对话、正确答案、奖励、个人解决方案、对话的标志位、对话轮次、对话内容以及对话角色等信息。数据集按照难度等级和类型进行分类，并提供了预测和个人提示对话内容。数据集分为训练集，共有339,308个示例。

This dataset contains conversation and solution information, suitable for text generation, dialogue systems, and other fields. It includes features such as index, prompt, whether it is the first round of conversation, ground truth, rewards, personal solutions, flags for conversations, turn of conversation, conversation content, and roles in the conversation. The dataset is categorized by difficulty level and type, and provides predicted and personal prompt conversation content. The dataset is split into a training set with a total of 339,308 examples.