Skip to content

Dataset · Rapidata

Human preference evaluations for Marey video generation

Rapidata/text-2-video-human-preferences-moonvalley-marey

1,313 examples with 74.8 thousand human ratings for videos generated by the Marey model.

What's inside

Provided in parquet format, this dataset comprises 1,313 rows containing videos, images, texts, and English tabular data that accumulated 74,844 ratings from roughly 15 thousand annotators via the Rapidata API. The data was gathered to evaluate Marey video generation quality on benchmarks.

What you can do with it

  • Training reward models for RLHF and aligning text-to-video generators with human preferences
  • Automated evaluation and quality comparison of video generation models
Tasks
Video classification, Text → video, Text classification
Languages
English
Size
1K–10K rows
Downloads this month
1.2K
Likes
9

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("Rapidata/text-2-video-human-preferences-moonvalley-marey")