Text-conditioned 3D molecule generation via E(3)-equivariant latent diffusion. A latent diffusion framework that generates physically consistent 3D molecular structures from natural language using an equivariant latent space and cross-attention conditioning.
drug-discovery molecular-generation egnn latent-diffusion text-guided-generation equivariant-graph-neural-network text-to-molecule
-
Updated
May 16, 2026 - Python