neko-llm/dna_dpo_hh-rlhf
收藏官方服务：
资源简介：
该数据集包含问题及其相关的输出，分为首选输出和非首选输出。每个问题都标注了可能的风险区域、伤害类型和具体伤害。数据集共有49027个训练示例。
The dataset includes questions and their associated outputs, divided into preferred and non-preferred outputs. Each question is labeled with potential risk areas, types of harm, and specific harms. The dataset consists of 49,027 training examples.
提供机构：
neko-llm
