Skip to content

Dataset · KBlueLeaf

Compressed Danbooru 2023 images

KBlueLeaf/danbooru2023-webp-4Mpixel

Recompressed Danbooru 2023 version: cleaned images up to 4 MP for training generative models.

What's inside

A resized and re-encoded version of danbooru2023: non-images and cropped files removed, images resized to 4 MP. Last id is 7,832,883; updates merged into the 2000~2999.tar archive. Size ranges from 100M to 1B examples, language is en, modalities are image and text.

What you can do with it

  • Training text-to-image models
  • LoRA fine-tuning for image generation
  • Testing image-classification and zero-shot models
Tasks
Image classification, Zero shot image classification, Text → image
Languages
English
Size
100M–1B rows
Downloads this month
1.5K
Likes
83
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("KBlueLeaf/danbooru2023-webp-4Mpixel")