lambda/sd-pokemon-diffusers

1---2language: 3  - en4thumbnail: "https://s3.amazonaws.com/moonup/production/uploads/1663756797814-62bd5f951e22ec84279820e8.png"5tags:6- stable-diffusion7- stable-diffusion-diffusers8- text-to-image9datasets:10- lambdalabs/pokemon-blip-captions11---12 13__Stable Diffusion fine tuned on Pokémon by [Lambda Labs](https://lambdalabs.com/).__14 15Put in a text prompt and generate your own Pokémon character, no "prompt engineering" required!16 17If you want to find out how to train your own Stable Diffusion variants, see this [example](https://github.com/LambdaLabsML/examples/tree/main/stable-diffusion-finetuning) from Lambda Labs.18 19 20![image.png](https://s3.amazonaws.com/moonup/production/uploads/1663756797814-62bd5f951e22ec84279820e8.png)21 22> Girl with a pearl earring, Cute Obama creature, Donald Trump, Boris Johnson, Totoro, Hello Kitty23 24## Usage25 26```bash27!pip install diffusers==0.3.028!pip install transformers scipy ftfy29```30 31```python32import torch33from diffusers import StableDiffusionPipeline34from torch import autocast35 36pipe = StableDiffusionPipeline.from_pretrained("lambdalabs/sd-pokemon-diffusers", torch_dtype=torch.float16)  37pipe = pipe.to("cuda")38 39prompt = "Yoda"40scale = 1041n_samples = 442 43# Sometimes the nsfw checker is confused by the Pokémon images, you can disable44# it at your own risk here45disable_safety = False46 47if disable_safety:48  def null_safety(images, **kwargs):49      return images, False50  pipe.safety_checker = null_safety51 52with autocast("cuda"):53  images = pipe(n_samples*[prompt], guidance_scale=scale).images54 55for idx, im in enumerate(images):56  im.save(f"{idx:06}.png")57```58 59## Model description60 61Trained on [BLIP captioned Pokémon images](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) using 2xA6000 GPUs on [Lambda GPU Cloud](https://lambdalabs.com/service/gpu-cloud) for around 15,000 step (about 6 hours, at a cost of about $10).62 63## Links64 65- [Lambda Diffusers](https://github.com/LambdaLabsML/lambda-diffusers)66- [Captioned Pokémon dataset](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions)67- [Model weights in Diffusers format](https://huggingface.co/lambdalabs/sd-pokemon-diffusers)68- [Original model weights](https://huggingface.co/justinpinkney/pokemon-stable-diffusion)69- [Training code](https://github.com/justinpinkney/stable-diffusion)70 71Trained by [Justin Pinkney](justinpinkney.com) ([@Buntworthy](https://twitter.com/Buntworthy)) at [Lambda Labs](https://lambdalabs.com/).