metadata
license: apache-2.0
library_name: diffusers
pipeline_tag: text-to-image
base_model:
- black-forest-labs/FLUX.1-dev
Scale-wise Distillation FLUX
Scale-wise Distillation (SwD) is a novel framework for accelerating diffusion models (DMs)
by progressively increasing spatial resolution during the generation process.
SwD achieves significant speedups (2.5× to 10×) compared to full-resolution models
while maintaining or even improving image quality.
Project page: https://yandex-research.github.io/swd
GitHub: https://github.com/yandex-research/swd
Demo: https://huggingface.co/spaces/dbaranchuk/Scale-wise-Distillation
Usage
Upgrade to the latest version of the 🧨 diffusers and 🧨 peft
pip install -U diffusers
pip install -U peft
and then you can run
import torch
from diffusers import FluxPipeline
from peft import PeftModel
pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev",
torch_dtype=torch.float16,
custom_pipeline="quickjkee/swd_pipeline_flux").to("cuda")
lora_path = "yresearch/swd_flux"
pipe.transformer = PeftModel.from_pretrained(
pipe.transformer,
lora_path,
)
sigmas = [1.0000, 0.8956, 0.7363, 0.6007, 0.0000]
scales = [64, 80, 96, 128]
prompt = "Cute winter dragon baby, kawaii, Pixar, ultra detailed, glacial background, extremely realistic."
image = pipe(
prompt=prompt,
height=int(scales[0] * 8),
width=int(scales[0] * 8),
scales=scales,
sigmas=sigmas,
timesteps=torch.tensor(sigmas[:-1], device="cuda") * 1000,
guidance_scale=4.5,
max_sequence_length=512,
).images[0]
Citation
@inproceedings{
starodubcev2026scalewise,
title={Scale-wise Distillation of Diffusion Models},
author={Nikita Starodubcev and Ilya Drobyshevskiy and Denis Kuznedelev and Artem Babenko and Dmitry Baranchuk},
booktitle={The Fourteenth International Conference on Learning Representations},
year={2026},
url={https://openreview.net/forum?id=Z06LNjqU1g}
}