ED-DiT pretrains a diffusion transformer on electron-density point clouds with a physical electron-number constraint, and the resulting encoder outperforms scratch models across six molecular tasks.
3d-molt5: leveraging discrete structural information for molecule-text modeling
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
ED-DiT: Physics-Guided Diffusion Pretraining for Transferable Molecular Representations from Electron Density
ED-DiT pretrains a diffusion transformer on electron-density point clouds with a physical electron-number constraint, and the resulting encoder outperforms scratch models across six molecular tasks.