ljw13/retro-star-training-datasets
收藏Hugging Face2025-10-16 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/ljw13/retro-star-training-datasets
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是为了优化大型语言模型在推理密集型文档检索任务上的表现而设计的。它包含文本排名和文本检索两种任务类型的数据，数据语言为英语，规模在10K到100K之间。数据集分为名为bright的配置，包含了两个数据分割：sft和rl。
The dataset is designed to optimize large language models for reasoning-intensive document retrieval tasks. It includes data for both text ranking and text retrieval tasks, with the language being English and the size ranging from 10K to 100K. The dataset is split into a configuration named bright, which contains two data splits: sft and rl.
提供机构：
ljw13
