laurentiubp/wizard-scored

该数据集包含消息列表、来源、评分和聚合评分等特征。消息列表包含内容和角色两个字段，评分部分包含连贯性、复杂性、正确性、帮助性、指令遵循、安全性、冗长性和冗长系数等多个子评分项。数据集分为训练集和测试集，训练集包含141570个样本，测试集包含1430个样本。

The dataset includes features such as a list of messages, source, scores, and an aggregate score. The list of messages contains fields for content and role. The scores section includes multiple sub-scores such as coherence, complexity, correctness, helpfulness, instruction-following, safety, verbosity, and verbosity coefficient. The dataset is divided into a training set with 141,570 samples and a test set with 1,430 samples.