Skip to content

Dataset · ProGamerGov

Regularization images for DreamBooth SD 1.5

ProGamerGov/StableDiffusion-v1-5-Regularization-Images

Class image set for prior preservation loss in DreamBooth when fine-tuning Stable Diffusion v1-5.

What's inside

A collection of regularization / class instance images for Stable Diffusion v1-5 designed for DreamBooth prior preservation loss. Files marked 'mse vae' are created with the stabilityai/sd-vae-ft-mse VAE. The data is provided in ZIP archives containing 512x512 PNGs; the image count is specified at the end of the filename. Size ranges from 1K to 10K images.

What you can do with it

  • DreamBooth fine-tuning of SD 1.5 with prior preservation
  • Comparing VAEs for regularization
  • Preparing LoRA / personalization without class forgetting
Size
1K–10K rows
Downloads this month
523
Likes
173

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("ProGamerGov/StableDiffusion-v1-5-Regularization-Images")