Dataset · inclusionAI
12 million instruction-based image editing examples
inclusionAI/ConceptEdit-12M
Large corpus of before-and-after pairs with English and Chinese text instructions and quality checks.
What's inside
A massive dataset (10 to 100 million records) for text-based image editing, packaged in .tar archives. Each example consists of an original image, an edited version, and a JSON file with instructions in English and Chinese, editing category, and VQA quality checks.
What you can do with it
- Train image-to-image editing models via text commands
- Train LoRA adapters for targeted visual edits
- Train VLM critics to verify how accurately an image edit was performed
- Tasks
- Image → image
- Languages
- English, Chinese
- Size
- 10M–100M rows
- Downloads this month
- 17K
- Likes
- 59
How to load it
from datasets import load_dataset
ds = load_dataset("inclusionAI/ConceptEdit-12M")