Skip to content

Dataset · Lixsp11

Sekai video dataset for world generation and research

Lixsp11/Sekai

Hundreds of thousands of videos with English captions from Shanghai AI Lab for video models.

What's inside

The repository features a dataset of 100,000 to 1 million examples, including video recordings and English text descriptions. Prepared by researchers from the Shanghai AI Laboratory and other scientific institutes for virtual and real-world space research tasks.

What you can do with it

  • Training generative text-to-video and image-to-video models
  • Developing environment simulation and navigation systems (World Models)
Tasks
Text → video, Image → video, Video → video
Languages
English
Size
100K–1M rows
Downloads this month
2.8K
Likes
46

How to load it

from datasets import load_dataset

ds = load_dataset("Lixsp11/Sekai")