Kyleyee/train_data_SFT_HH
收藏官方服务：
资源简介：
HH-RLHF-Helpful-Base数据集是一个经过处理的Anthropic HH-RLHF数据集版本，专门用于通过TRL库进行偏好学习和对齐任务的模型训练。该数据集包含基于人类评估者对响应帮助性的偏好而标记为“选中”或“拒绝”的对话格式文本样本对。
The HH-RLHF-Helpful-Base dataset is a processed version of Anthropics HH-RLHF dataset, specifically curated for model training using the TRL library for preference learning and alignment tasks. It includes conversational text sample pairs labeled as chosen or rejected based on human evaluators preferences regarding the helpfulness of responses.
提供机构：
Kyleyee
