MWilinski/hh-rlhf-helpful-rejection-sampled
收藏Hugging Face2025-11-05 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/MWilinski/hh-rlhf-helpful-rejection-sampled
下载链接
链接失效反馈官方服务：
资源简介：
HH-RLHF帮助性拒绝采样数据集是一个处理过的Anthropic HH-RLHF数据目录变体，它将原始字符串转换为与TRL偏好和RM训练管道兼容的对话聊天格式。这个转换保留了人类和助手的话轮，并且明确地区分了所选和拒绝的延续。
HH-RLHF Helpful Rejection Sampled dataset is a processed variant of the Anthropic HH-RLHF data directory, which converts the raw strings into a conversational chat format compatible with the TRL preference and RM training pipelines. The transformation preserves the human and assistant turns and clearly separates the chosen and rejected continuations.
提供机构：
MWilinski
