udkai/alignment
收藏官方服务：
资源简介：
AI价值对齐数据集包含了用于对齐小型指令调整语言模型与道德和伦理价值结构的实验资源和数据。该数据集基于一篇研究论文，使用了名为Moral Ordinals的新颖方法来对抽象价值进行排序，并测试它们在AI对齐中的作用。
The AI Value Alignment Dataset includes experimental resources and data for aligning small instruction-tuned language models with moral and ethical value structures. The dataset is based on a research paper that employs a novel approach called Moral Ordinals to order abstract values and test their influence in AI alignment.
提供机构：
udkai
