Lakshan2003/gemini-2.5-flash-customerservice-Human-eval-data

这个数据集包含了对话信息，其中包括会话ID、会话阶段、对话指令、历史对话记录、历史对话摘要、用户提出的问题、问题的真实答案、系统生成的答案，以及针对生成答案的几个评估指标，如人类相似度、连贯性与上下文理解能力、语调与清晰度、任务的适宜性。数据集分为训练集，共有500个示例，数据大小为1,456,398字节。

This dataset contains conversational information, including conversation ID, conversation stage, dialogue instructions, historical dialogue records, summary of historical dialogues, questions raised by users, true answers to the questions, answers generated by the system, and several evaluation metrics for the generated answers, such as human likeness, coherence and context understanding, tone and clarity, and task appropriateness. The dataset is split into a training set with a total of 500 examples and a data size of 1,456,398 bytes.