ML-GSAI/Scaling-Diffusion-Transformers-muP

[NeurIPS 2025] Official implementation for our paper "Scaling Diffusion Transformers Efficiently via μP".

33
/ 100
Emerging

This project offers a method for efficiently scaling Diffusion Transformers, which are powerful AI models used for generating images and other media. It helps AI researchers and engineers develop and deploy larger, more capable generative AI models by significantly reducing the time and computational resources needed to fine-tune their hyperparameters. Researchers can input smaller model configurations and receive optimized hyperparameters that transfer directly to much larger models, speeding up the development of high-quality image generation systems.

Use this if you are developing or training large-scale diffusion models (like those for image generation) and want to drastically cut down on the computational cost and time spent on hyperparameter tuning.

Not ideal if you are working with smaller models, non-generative AI tasks, or if you are not deeply involved in the research and development of large-scale AI models.

generative-AI diffusion-models AI-model-scaling hyperparameter-optimization image-synthesis
No Package No Dependents
Maintenance 6 / 25
Adoption 9 / 25
Maturity 15 / 25
Community 3 / 25

How are scores calculated?

Stars

95

Forks

1

Language

Python

License

MIT

Last pushed

Nov 02, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/ML-GSAI/Scaling-Diffusion-Transformers-muP"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.