singhalrk/combined-wildchat-qwen3-8b-responses

这是一个多语言对话数据集，包含超过100万条对话示例。数据集记录了用户查询和模型响应，并涉及毒性检测、意图分类和描述生成等任务。特征包括对话哈希、时间戳、语言、毒性标签、国家、IP哈希、查询内容、模型原始输出、响应、思考过程以及完成原因等。数据用于训练和评估对话模型，支持多语言环境下的安全性和性能分析。

This is a multilingual dialogue dataset containing over 1 million conversation examples. It records user queries and model responses, and involves tasks such as toxicity detection, intent classification, and description generation. Features include conversation hash, timestamp, language, toxicity label, country, hashed IP, query content, model raw output, response, thinking process, and finish reason. The data is used for training and evaluating dialogue models, supporting safety and performance analysis in multilingual environments.