README.md

---
language: 
  - en
thumbnail: "https://staticassetbucket.s3.us-west-1.amazonaws.com/avatar_grid.png"
tags:
- dreambooth
- stable-diffusion
- stable-diffusion-diffusers
- text-to-image
---

# Dreambooth style: Avatar

__Dreambooth finetuning of Stable Diffusion (v1.5.1) on Avatar art style by [Lambda Labs](https://lambdalabs.com/).__

## About

Put in a text prompt and generate your own Avatar style image!

If you want to find out how to train your own Dreambooth Style, see this example (link lambda blog)
// ![pk1.jpg](https://staticassetbucket.s3.us-west-1.amazonaws.com/avatar_grid.png)
> descriptions?

## Usage

To run model locally:
```bash
pip install accelerate torchvision transformers>=4.21.0 ftfy tensorboard modelcards
```

```python
import torch
from diffusers import StableDiffusionPipeline
from torch import autocast

pipe = StableDiffusionPipeline.from_pretrained("lambdalabs/dreambooth-avatar", torch_dtype=torch.float16)  
pipe = pipe.to("cuda")

prompt = "Yoda, avatarart style person"
scale = 7.5
n_samples = 4

# Sometimes the nsfw checker is confused by the Naruto images, you can disable
# it at your own risk here
disable_safety = False

if disable_safety:
  def null_safety(images, **kwargs):
      return images, False
  pipe.safety_checker = null_safety

with autocast("cuda"):
  images = pipe(n_samples*[prompt], guidance_scale=scale).images

for idx, im in enumerate(images):
  im.save(f"{idx:06}.png")
```

## Model description

Trained on 512x512 Avatar character images using 2xA6000 GPUs on [Lambda GPU Cloud](https://lambdalabs.com/service/gpu-cloud) for around 30,000 step (about 1 hours, at a cost of about $2).

## Links


- [Lambda Diffusers](https://github.com/LambdaLabsML/lambda-diffusers)
- [Model weights in Diffusers format](https://huggingface.co/lambdalabs/sd-naruto-diffusers)
- [Naruto diffusers repo](https://github.com/eolecvk/naruto-sd)

Trained by Eole Cervenka