This pipeline was finetuned from CompVis/stable-diffusion-v1-4 on the middles/kanjiv1 dataset. Below are some example images generated with the finetuned pipeline using the following prompts: ['=', 'Japanese Kanji']:

Pipeline usage

You can use the pipeline like so:

python

from diffusers import DiffusionPipeline
import torch

pipeline = DiffusionPipeline.from_pretrained("middles/kanji_demo", torch_dtype=torch.float16)
prompt = "="
image = pipeline(prompt).images[0]
image.save("my_image.png")

Training info

These are the key hyperparameters used during training:

Epochs: 13
Learning rate: 1e-05
Batch size: 1
Gradient accumulation steps: 4
Image resolution: 512
Mixed-precision: fp16

More information on all the CLI arguments and the environment are available on your wandb run page.