|
--- |
|
title: README |
|
emoji: 🚀 |
|
colorFrom: red |
|
colorTo: purple |
|
sdk: static |
|
pinned: false |
|
--- |
|
<div class="prose prose-sm dark:prose-light"> |
|
<p> |
|
Welcome to CompVis! |
|
</p> |
|
|
|
<p> |
|
We host public weights for Latent Diffusion and Stable Diffusion models. There are several options to choose from, please check the details below. |
|
</p> |
|
|
|
|
|
<p> |
|
<b>Stable Diffusion Models</b> |
|
</p> |
|
|
|
<p> |
|
Stable Diffusion is a latent text-to-image diffusion model capable of generating photo-realistic images given any text input. For more information about how Stable Diffusion works, please have a look at <a href="https://hf.co/blog/stable_diffusion">🤗's Stable Diffusion with 🧨 Diffusers blog</a>. |
|
</p> |
|
|
|
<p> |
|
We recommend you use Stable Diffusion with <a href="https://github.com/huggingface/diffusers">🤗 Diffusers library</a>. You can also use the original <a href="https://github.com/compvis/stable-diffusion">CompVis code</a>. There are variants of the weights depending on: |
|
<ul> |
|
<li>The library they are intended for.</li> |
|
<li>The training regime. There are 4 training versions: <tt>v1-1</tt> through <tt>v1-4</tt>. Each one was created from the checkpoint of the previous version, and was trained for additional steps in specific variants of the dataset. |
|
</p> |
|
|
|
<p> |
|
Please, refer to the details in the following table to choose the weights appropriate for your use. |
|
</p> |
|
|
|
| Model | Library | Details | |
|
|-------------------------------------------------------------------------------------------------|---------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------| |
|
| [stable-diffusion-v1-1](https://huggingface.co/CompVis/stable-diffusion-v1-1) | 🤗 [Diffusers](https://github.com/huggingface/diffusers) | 237k steps at resolution 256x256 on laion2B-en. <br> 194k steps at resolution 512x512 on laion-high-resolution. | |
|
| [stable-diffusion-v1-2](https://huggingface.co/CompVis/stable-diffusion-v1-2) | 🤗 [Diffusers](https://github.com/huggingface/diffusers) | v1-1 plus: <br> 515k steps at 512x512 on "laion-improved-aesthetics". | |
|
| [stable-diffusion-v1-3](https://huggingface.co/CompVis/stable-diffusion-v1-3) | 🤗 [Diffusers](https://github.com/huggingface/diffusers) | v1-2 plus: <br> 195k steps at 512x512 on "laion-improved-aesthetics", <br> with 10% dropping of text-conditioning. | |
|
| [stable-diffusion-v1-4](https://huggingface.co/CompVis/stable-diffusion-v1-4) | 🤗 [Diffusers](https://github.com/huggingface/diffusers) | v1-2 plus: <br> 225k steps at 512x512 on "laion-aesthetics v2 5+", <br>with 10% dropping of text conditioning. | |
|
| [stable-diffusion-v-1-1-original](https://huggingface.co/CompVis/stable-diffusion-v-1-1-original) | CompVis | 237k steps at resolution 256x256 on laion2B-en. <br> 194k steps at resolution 512x512 on laion-high-resolution. | |
|
| [stable-diffusion-v-1-2-original](https://huggingface.co/CompVis/stable-diffusion-v-1-2-original) | CompVis | v1-1 plus: <br> 515k steps at 512x512 on "laion-improved-aesthetics". | |
|
| [stable-diffusion-v-1-3-original](https://huggingface.co/CompVis/stable-diffusion-v-1-3-original) | CompVis | v1-2 plus: <br> 195k steps at 512x512 on "laion-improved-aesthetics", <br> with 10% dropping of text-conditioning. | |
|
| [stable-diffusion-v-1-4-original](https://huggingface.co/CompVis/stable-diffusion-v-1-4-original) | CompVis | v1-2 plus: <br> 225k steps at 512x512 on "laion-aesthetics v2 5+", <br> with 10% dropping of text conditioning. | |
|
</div> |
