MinkaiXu/GeoDiff
A geometric diffusion model using graph neural networks to generate 3D molecular conformations.
Not currently ranked — collecting fresh signals.
star history
GeoDiff implements a score-based diffusion model for generating stable molecular 3D structures from 2D molecular graphs. It uses equivariant graph neural networks to learn score functions in the diffusion process, enabling unconditional generation of molecular conformations for drug discovery and computational chemistry applications.
Frequently asked
- What is MinkaiXu/GeoDiff?
- A geometric diffusion model using graph neural networks to generate 3D molecular conformations.
- Is GeoDiff open source?
- Yes — MinkaiXu/GeoDiff is open source, released under the MIT license.
- What language is GeoDiff written in?
- MinkaiXu/GeoDiff is primarily written in Python.
- How popular is GeoDiff?
- MinkaiXu/GeoDiff has 415 stars on GitHub.
- Where can I find GeoDiff?
- MinkaiXu/GeoDiff is on GitHub at https://github.com/MinkaiXu/GeoDiff.