arranonymsub/HiCUPID
收藏Hugging Face2025-02-09 更新2024-12-21 收录
下载链接：
https://hf-mirror.com/datasets/arranonymsub/HiCUPID
下载链接
链接失效反馈官方服务：
资源简介：
HiCUPID（对话用户档案包容性数据集）是一个合成的对话数据集，旨在训练和评估大型语言模型（LLMs）在个性化性能方面的表现。HiCUPID包含600个独特的用户档案（每个档案包含约15,000个标记的对话历史）和24,000个针对每个用户的问题-答案对。数据集分为三个子集：Dialogue、QA和Evaluation。Dialogue子集包含用户与助手之间的对话历史，QA子集包含与每个用户相关的问题-答案对，Evaluation子集包含评估结果。
HiCUPID (Conversational User Profile Inclusive Dataset) is a synthetic dialogue dataset designed to train and evaluate large language models (LLMs) in terms of personalization performance. HiCUPID consists of 600 unique user profiles (dialogue histories, each consisting of ~15,000 tokens), and 24,000 question-answer pairs specific to each user. The dataset is divided into three subsets: Dialogue, QA, and Evaluation. The Dialogue subset contains dialogue histories between users and an assistant, the QA subset contains question-answer pairs corresponding to each user, and the Evaluation subset contains evaluation results.
提供机构：
arranonymsub
