ontocord/VALID
收藏Hugging Face2025-09-27 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/ontocord/VALID
下载链接
链接失效反馈官方服务：
资源简介：
VALID（视频-音频大交错数据集）是一个由大约72万个来自YouTube的知识共享许可视频组成的多模态数据集，经过处理后形成了音视频文本数据记录，用于机器学习研究。该数据集包括图像、音频片段和文本，旨在帮助模型理解不同模态之间的关系，适用于多模态表征学习等应用。
VALID (Video-Audio Large Interleaved Dataset) is a multimodal dataset composed of approximately 720,000 Creative Commons licensed videos from YouTube, processed into audio-video-text data records for machine learning research. The dataset includes images, audio snippets, and text, designed to help models understand the relationships between different modalities, suitable for applications such as multimodal representation learning.
提供机构：
ontocord
