MQDialog
收藏资源简介:
MQDialog数据集由上海交通大学和阿里巴巴集团共同构建,旨在为个性化语言模型生成提供基准测试。该数据集包含173个提问者和12个回答者的对话记录,涵盖英语和中文脚本以及微信记录。数据集的创建过程包括从多种来源提取和清理对话数据,并通过聚类相似问题来优化对比学习。MQDialog数据集主要应用于个性化语言模型的研究,旨在解决不同用户对相同查询生成定制化响应的问题。
The MQDialog dataset was co-developed by Shanghai Jiao Tong University and Alibaba Group, with the objective of providing a benchmark for personalized language model generation. This dataset includes 173 conversation records involving 173 questioners and 12 responders, covering English and Chinese scripts as well as WeChat conversation logs. The development process of the MQDialog dataset entails extracting and cleaning conversation data from multiple sources, and optimizing contrastive learning via clustering similar questions. The MQDialog dataset is mainly applied to research on personalized language models, aiming to solve the problem of generating customized responses for different users when facing the same query.

- 1Personalized LLM for Generating Customized Responses to the Same Query from Different Users上海交通大学 · 2024年



