ultrafeedback
收藏arXiv2025-09-30 收录
下载链接：
https://huggingface.co/datasets/argilla/ultrafeedback-binarized-preferences-cleaned
下载链接
链接失效反馈官方服务：
资源简介：
该数据集被用于强化学习领域，特别是从人类反馈中训练模型，它不仅用于训练奖励模型，还用于评估对齐性能。具体任务为强化学习从人类反馈（Rlhf）。
This dataset is applied in the field of reinforcement learning, particularly for training models with human feedback. It is used not only to train reward models but also to evaluate alignment performance. The specific task is Reinforcement Learning from Human Feedback (RLHF).
提供机构：
Hugging Face
