Skip to content

Dataset · bigfacing

Million-scale dataset for instruction-based video editing

bigfacing/GOKU-2M

1 to 10 million before-and-after video pairs with English text instructions for 10 editing tasks.

What's inside

A large unified corpus and benchmark for instruction-based video editing in English, covering 10 types of edits. Each example consists of an input video, an edited target video, and one or more text instructions describing the modification; due to the massive volume, the data is split across two repositories.

What you can do with it

  • Training video-to-video editing models based on text prompts
  • Evaluating generative video models across 10 different editing scenarios
  • LoRA fine-tuning for local object or style modification in video clips
Tasks
Text → video, Video → video
Languages
English
Size
1M–10M rows
Downloads this month
1.9K
Likes
54
Access
on request on Hugging Face

How to load it

from datasets import load_dataset

ds = load_dataset("bigfacing/GOKU-2M")