ShareGPT_Vicuna_unfiltered
收藏arXiv2025-09-30 收录
下载链接：
https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是用于训练Vicuna模型的ShareGPT数据集的再现版本，它包含了没有对应回答示范的提示。此外，该数据集还用于与LLaMA-2模型对齐，规模为9万个提示，任务是基于提示的语言模型对齐训练。
This dataset is a reproduced iteration of the ShareGPT dataset intended for Vicuna model training, which includes prompts that lack corresponding response demonstrations. Furthermore, this dataset is utilized for alignment with the LLaMA-2 model, comprising 90,000 prompts, and is designed for prompt-based language model alignment training.
提供机构：
Hugging Face
