Hugging Face's logo

nroggendorff
/
anime

Unconditional Image Generation
Diffusers
TensorBoard
Safetensors
DDPMPipeline
diffusion
Model card Files Metrics Community
anime / README.md
nroggendorff's picture
nroggendorff
Update README.md
d46db24 verified
preview code
|
raw
history blame contribute delete
1.64 kB
metadata
license: mit
metrics:
  - mse
library_name: diffusers
tags:
  - diffusion
pipeline_tag: unconditional-image-generation

Anime Face Model Card

DDPMAnimes is a latent noise-to-image diffusion model capable of generating images of Anime girls (no I'm not weird). For more information about how Stable Diffusion functions, please have a look at 🤗's Stable Diffusion blog.

You can use this with the 🧨Diffusers library from Hugging Face.

So cute, right?

Diffusers 

from diffusers import DiffusionPipeline

pipeline = DiffusionPipeline.from_pretrained("nroggendorff/anime")
pipe = pipeline.to("cuda")

image = pipe().images[0]  

image.save("anime_face.png")

Model Details

  • train_batch_size: 16
  • eval_batch_size: 16
  • num_epochs: 50
  • gradient_accumulation_steps: 1
  • learning_rate: 1e-4
  • lr_warmup_steps: 500
  • mixed_precision: "fp16"
  • eval_metric: "mean_squared_error"

Bias

  • This model may exhibit biases due to its training data. It will not display images of overweight or otherwise non-traditionally attractive women, as it is sexist.

Limitations

  • The model does not achieve perfect photorealism
  • The model cannot render legible text
  • The model was trained on a medium-to-large-scale dataset: huggan/few-shot-anime-face

Developed by

  • Noa Linden Roggendorff

This model card was written by Noa Roggendorff and is based on the Stable Diffusion v1-5 Model Card.