Skip to content

Dataset · TIGER-Lab

OmniEdit: 1.2M Image Editing Pairs

TIGER-Lab/OmniEdit-Filtered-1.2M

1.2 million image-text pairs for seven editing tasks, filtered via GPT-4o scores.

What's inside

Contains images and text for seven distinct image editing tasks across arbitrary aspect ratios. Examples were selected based on importance scores from large multimodal models like GPT-4o rather than CLIP-score. English language, sized 1M–10M.

What you can do with it

  • Training image editing models
  • LoRA for multimodal editing
  • Evaluating filtered pair quality
Languages
English
Size
1M–10M rows
Downloads this month
58.2K
Likes
139

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("TIGER-Lab/OmniEdit-Filtered-1.2M")