LangAGI-Lab/llama-8b-verified-7k-dpo-preference-set
收藏Hugging Face2025-02-11 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/LangAGI-Lab/llama-8b-verified-7k-dpo-preference-set
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含指令、答案以及被选中和被拒绝的回答的相关信息。具体字段包括指令文本、答案文本、选中答案的正确性、选中答案的文本、选中答案的字符长度、被拒绝答案的正确性、被拒绝答案的文本、被拒绝答案的字符长度、最小思考令牌数量以及数据来源。数据集分为训练集，包含7422个示例，大小为89955487字节。
The dataset includes instructions, answers, and information about the chosen and rejected responses. Specific fields include instruction text, answer text, correctness of the chosen answer, text of the chosen answer, token length of the chosen answer, correctness of the rejected answer, text of the rejected answer, token length of the rejected answer, number of mini thought tokens, and data source. The dataset is split into a training set, which contains 7422 examples and is 89955487 bytes in size.
提供机构：
LangAGI-Lab
