MinkaiXu/GeoLDM
Geometric Latent Diffusion Models for 3D Molecule Generation
Combines a trainable variational autoencoder with equivariant diffusion in latent space to generate valid 3D molecular geometries while preserving SE(3) symmetries. Supports conditional generation of molecules with target properties (e.g., polarizability, HOMO-LUMO gap) using property classifiers for evaluation. Built on equivariant graph neural networks (EGNN) and includes pretrained models for QM9 and GEOM drug datasets.
273 stars. No commits in the last 6 months.
Stars
273
Forks
49
Language
Python
License
MIT
Category
Last pushed
Jun 09, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/MinkaiXu/GeoLDM"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
MinkaiXu/GeoDiff
Implementation of GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generation (ICLR 2022).
caio-freitas/GraphARM
An implementation of the Autoregressive Diffusion Model for Graph Generation from [Kong et al. 2023]
microsoft/foldingdiff
Diffusion models of protein structure; trigonometry and attention are all you need!
aqlaboratory/genie
De Novo Protein Design by Equivariantly Diffusing Oriented Residue Clouds
pengxingang/MolDiff
MolDiff: Addressing the Atom-Bond Inconsistency Problem in 3D Molecule Diffusion Generation