LLaVA-CoT-100K
收藏arXiv2025-09-30 收录
下载链接：
https://huggingface.co/datasets/xkev/llava-cot-100k
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为LLaVA-Instruct，包含了英文版和经过翻译及筛选的阿拉伯语版。该数据集在Dallah的微调阶段被使用，规模达到了15万样本，主要用于阿拉伯语的预训练模型微调任务。
This dataset is named LLaVA-Instruct, which includes an English version and an Arabic version that has been translated and filtered. It was used during the fine-tuning stage of Dallah, with a total of 150,000 samples, and is mainly applied to fine-tuning tasks for Arabic pre-trained models.
