Kyleyee/train_data_HH_sft
收藏Hugging Face2025-03-14 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/Kyleyee/train_data_HH_sft
下载链接
链接失效反馈官方服务：
资源简介：
HH-RLHF-Helpful-Base数据集是一个专门为使用TRL库进行偏好学习和对齐任务而加工的Anthropic HH-RLHF数据集的版本。它包含了成对的文本样本，每个样本都被人类评估者标记为“选中”或“拒绝”，以表示响应的有帮助性。这个数据集可以帮助模型学习在生成响应时的人类偏好，从而更有效地帮助用户。
The HH-RLHF-Helpful-Base dataset is a curated version of Anthropics HH-RLHF dataset, specifically processed for training models on preference learning and alignment tasks using the TRL library. It includes pairs of text samples, each labeled as chosen or rejected based on human evaluators preferences for helpfulness. This dataset enables models to learn human preferences in generating helpful responses, enhancing their effectiveness in assisting users.
提供机构：
Kyleyee
