News list for "ontocord"

Grass, Ontocord, and LAION jointly release the first video-audio interleaved dataset, VALID.

AI industry leaders Grass, Ontocord, and LAION have announced the joint release of the VALID (Video-Audio Large Interleaved Dataset) dataset. Built on the Grass video repository, the dataset contains 30 million audio clips that are interleaved with images and text, making it the industry's first video-audio interleaved dataset. The release of VALID will provide new data support for the training of multimodal AI models.

clock
2024-12-06 10:39:41
Grass、Ontocord和LAION联合发布首个视频-音频交错数据集VALID

AI领域知名机构Grass、Ontocord和LAION宣布联合发布VALID(Video-Audio Large Interleaved Dataset)数据集。 该数据集基于Grass视频仓库构建,包含3000万条音频片段,这些音频片段与图像和文本进行了交错配对,是业内首个视频-音频交错数据集。VALID的发布将为多模态AI模型的训练提供新的数据支持。

clock
2024-12-06 10:39:41