Alrightlone/OBS-Diff
[ICLR 2026] Offical implementation of "OBS-Diff".
This project helps machine learning engineers and researchers reduce the computational cost of large text-to-image diffusion models. By taking an existing diffusion model and a calibration dataset, it produces a smaller, more efficient version of the model that generates high-quality images faster. This is ideal for those who deploy or research advanced image generation models and need to optimize their performance.
Use this if you need to compress large text-to-image diffusion models like Stable Diffusion to reduce their memory footprint and accelerate image generation without significantly sacrificing visual quality.
Not ideal if you are working with non-diffusion models or require extreme precision where even minimal visual degradation is unacceptable.
Stars
54
Forks
2
Language
Python
License
—
Category
Last pushed
Mar 05, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/Alrightlone/OBS-Diff"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
xie-lab-ml/Golden-Noise-for-Diffusion-Models
[ICCV2025] The code of our work "Golden Noise for Diffusion Models: A Learning Framework".
yulewang97/ERDiff
[NeurIPS 2023 Spotlight] Official Repo for "Extraction and Recovery of Dpatio-temporal Structure...
UNIC-Lab/RadioDiff
This is the code for the paper "RadioDiff: An Effective Generative Diffusion Model for...
pantheon5100/pid_diffusion
This repository is the official implementation of the paper: Physics Informed Distillation for...
zju-pi/diff-sampler
An open-source toolbox for fast sampling of diffusion models. Official implementations of our...