Conversational Task Assistant (CTA) Dialogues

该数据集包含了20,000条对话，这些对话基于1,000个独特的食谱生成，模拟了烹饪领域中的用户交互。为了提高泛化能力，这些对话在创建时采用了任务抽样方法，并确保不同分割之间任务不共享。每个对话平均包含9.6个回合。该数据集的规模为20,000条对话，其任务是在烹饪领域提供任务协助。

This dataset contains 20,000 dialogues generated from 1,000 unique recipes, simulating user interactions in the culinary domain. To enhance generalization performance, task sampling strategies were utilized during dialogue creation, and strict measures were taken to ensure that no tasks were shared across different data splits. On average, each dialogue consists of 9.6 conversational turns. The total size of this dataset is 20,000 dialogues, and its target is to provide task-oriented assistance in the culinary field.