Skip to content

Dataset · Lewandofski

Video editing instruction dataset

Lewandofski/OpenVE-3M

Large open dataset of video pairs with instructions for training video editing models.

What's inside

OpenVE-3M is a large-scale open dataset for instruction-guided video editing. The card notes it includes eight main categories, though the list is truncated in the available fragment. The modality is video, with a size between 1M and 10M examples. Language and file formats are not specified.

What you can do with it

  • Training instruction-guided video editing models
  • Fine-tuning video models or LoRA for text-based edits
  • Evaluating video editing systems
Size
1M–10M rows
Downloads this month
10.3K
Likes
98

How to load it

from datasets import load_dataset

ds = load_dataset("Lewandofski/OpenVE-3M")