DeepGlint-AI/DanQing100M
收藏官方服务：
资源简介：
DanQing是一个大规模的中文视觉语言预训练数据集，包含1亿个图像-文本对，数据来源于2024-2025年的网络数据，经过严格筛选，具有较高的数据质量。数据集主要用于零样本图像分类、图像到文本等任务。
DanQing is a large-scale Chinese vision-language pre-training dataset containing 100 million image-text pairs collected from Common Crawl. The data is primarily built from 2024-2025 web data and is curated through a rigorous selection process, yielding superior data quality. The dataset is designed for tasks such as zero-shot image classification and image-to-text.
提供机构：
DeepGlint-AI
