BSC-LT/ACAData
收藏Hugging Face2026-03-06 更新2025-10-18 收录
下载链接：
https://hf-mirror.com/datasets/BSC-LT/ACAData
下载链接
链接失效反馈官方服务：
资源简介：
ACAData是一个多语言指令微调数据集，包含来自学术领域的平行文本段落。数据集包括两个子集：ACAD-Train用于训练，ACAD-Bench用于基准测试。数据以原始Parquet格式提供，包括ISO语言代码和文本内容。数据集在CC BY 4.0许可下发布，旨在提高学术领域的机器翻译性能。
ACAData is a multilingual instruction tuning dataset containing parallel text paragraphs from the academic domain. It includes two subsets: ACAD-Train for training and ACAD-Bench for benchmarking. The data is provided in raw Parquet format and includes ISO language codes and textual content. The dataset is licensed under CC BY 4.0 and aims to improve machine translation performance in the academic domain.
提供机构：
BSC-LT
