metadata

license: other
tags:
  - stable-diffusion
  - text-to-image
inference: false

Stable Diffusion

Stable Diffusion is a latent text-to-image diffusion model capable of generating photo-realistic images given any text input. This model card gives an overview of all available model checkpoints. For more in-detail model cards, please have a look at the model repositories listed under Model Access.

** Stable Diffusion V1**

In its first version, 4 model checkpoints are released: stable-diffusion-v1-1, stable-diffusion-v1-2, stable-diffusion-v1-3 and stable-diffusion-v1-4. Higher versions have been trained for longer and are thus usually better in terms of image generation quality then lower versions. More specifically:

stable-diffusion-v1-1: The checkpoint is randomely initialized and has been trained on 237,000 steps at resolution 256x256 on laion2B-en. 194,000 steps at resolution 512x512 on laion-high-resolution (170M examples from LAION-5B with resolution >= 1024x1024).
stable-diffusion-v1-2 (https://huggingface.co/CompVis/stable-diffusion-v1-2): The checkpoint is resumed training from stable-diffusion-v1-1. 515,000 steps at resolution 512x512 on "laion-improved-aesthetics" (a subset of laion2B-en, filtered to images with an original size >= 512x512, estimated aesthetics score > 5.0, and an estimated watermark probability < 0.5. The watermark estimate is from the LAION-5B metadata, the aesthetics score is estimated using an improved aesthetics estimator).
stable-diffusion-v1-3 (https://huggingface.co/CompVis/stable-diffusion-v1-3): The checkpoint is resumed training from stable-diffusion-v1-2. 195,000 steps at resolution 512x512 on "laion-improved-aesthetics" and 10 % dropping of the text-conditioning to improve classifier-free guidance sampling
stable-diffusion-v1-4 (https://huggingface.co/CompVis/stable-diffusion-v1-4) The checkpoint is resumed training.

The model can be used both with 🤗's diffusers library or the original Stable Diffusion GitHub repository.

Model access

Each checkpoint can be accessed as soon as having "click-requested" them on the respective model repositories.

For 🤗's diffusers:

For with the original Stable Diffusion GitHub repository:

Citation

    @InProceedings{Rombach_2022_CVPR,
        author    = {Rombach, Robin and Blattmann, Andreas and Lorenz, Dominik and Esser, Patrick and Ommer, Bj\"orn},
        title     = {High-Resolution Image Synthesis With Latent Diffusion Models},
        booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
        month     = {June},
        year      = {2022},
        pages     = {10684-10695}
    }

This model card was written by: Robin Rombach and Patrick Esser and is based on the DALL-E Mini model card.