A distributed deep learning system designed for optimizing and implementing various distributed training technologies, integrating with community projects, and adapting to heterogeneous chips.

A video generation model focusing on controllable generation of visual content, including text-to-video, image-to-video, digital humans, and video editing.