AI4Science-WestlakeU/t_scend

This repo is the code for T-SCEND, a novel framework that significantly improves diffusion model’s reasoning capabilities with better energy-based training and scaling up test-time computation.

33
/ 100
Emerging

This project helps AI researchers and machine learning engineers significantly improve the reasoning capabilities of diffusion models. It takes existing diffusion models and through advanced training and scalable computation, outputs models that can solve complex tasks, like larger mazes or Sudoku puzzles, with higher accuracy even when trained on simpler versions. This is designed for those pushing the boundaries of AI problem-solving.

Use this if you are developing or experimenting with diffusion models and need to enhance their ability to reason and generalize to more complex problems than their training data.

Not ideal if you are looking for a ready-to-use application to solve specific problems without delving into model training and research.

AI Research Diffusion Models Machine Learning Engineering Problem Solving AI Generative AI
No Package No Dependents
Maintenance 6 / 25
Adoption 7 / 25
Maturity 16 / 25
Community 4 / 25

How are scores calculated?

Stars

26

Forks

1

Language

Python

License

MIT

Last pushed

Oct 19, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/AI4Science-WestlakeU/t_scend"

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