Locutusque/Wild-GPT-4-Turbo-Cleaned

该数据集是对allenai/WildChat-1M数据集进行过滤，仅保留GPT-4-Turbo生成的回答，并进行了去重、去噪和拒绝移除等处理，最终转换为ShareGPT格式。数据集包含两个配置（cleaned和default），每个配置都包含多个特征字段，如conversation_hash、model、timestamp等。数据集的分割信息为train，包含大量示例和字节数。

The dataset contains processed conversational data, primarily for analyzing and evaluating the quality and safety of conversation content. The dataset includes two configurations: cleaned and default, each with detailed feature descriptions such as conversation hash, model, timestamp, conversation content, country, language, whether it has been redacted, etc. The dataset also includes multiple classifications and scores for conversation content, such as whether it contains harassment, hate speech, violence, etc., as well as scores and flags for these contents. The dataset has undergone various processing steps, such as deduplication, deslopping, refusal removal, etc., and has been converted to the ShareGPT format.