hammh0a/Hala-4.6M-SFT
收藏官方服务:
资源简介:
Hala数据集是一个大规模的以阿拉伯语为中心的指令调整语料库,通过将广泛使用的英语指令数据集翻译成阿拉伯语构建而成,用于训练Hala系列模型,并旨在填补高质量阿拉伯语指令跟随资源的空白。
The Hala Dataset is a large-scale Arabic-centric instruction tuning corpus, built by translating widely-used English instruction datasets into Arabic, designed for training the Hala family of models and aims to fill the gap in high-quality Arabic instruction-following resources.
提供机构:
hammh0a


