rlhn/hn-remove-250K
收藏Hugging Face2025-05-27 更新2025-11-01 收录
下载链接：
https://hf-mirror.com/datasets/rlhn/hn-remove-250K
下载链接
链接失效反馈官方服务：
资源简介：
RLHN是一个级联LLM框架，用于准确重标记现有信息检索/生成训练数据集中的难负例。这个数据集（250K训练对）包含了BGE训练集合中7个数据集的查询、正例和难负例（去除错误的负例）。该数据集可用于微调嵌入、ColBERT或多元向量以及重排模型。
RLHN is a cascading LLM framework designed to accurately relabel hard negatives in existing IR/RAG training datasets. This dataset (250K training pairs) contains queries, positives, and hard negatives (with dropped false negatives) for 7 datasets in the BGE training collection. It can be used to fine-tune embeddings, ColBERT or multi-vector, and reranker models.
提供机构：
rlhn
