Dataset · saiyan-world
Goku-MovieGenBench video data
saiyan-world/Goku-MovieGenBench
Video data for training and evaluating the Goku text-to-video generative model.
What's inside
A video dataset associated with the Goku paper on streaming video generative models. It contains between 1,000 and 10,000 video records; the language and exact format are not specified in the card. It is intended for text-to-video tasks.
What you can do with it
- Train or fine-tune a text-to-video model.
- Evaluate video generation quality from text descriptions.
- Test a video generative model on the MovieGenBench benchmark.
- Tasks
- Text → video
- Size
- 1K–10K rows
- Downloads this month
- 1.2K
- Likes
- 225
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("saiyan-world/Goku-MovieGenBench")