Atom-JEPA: Joint-Embedding Predictive Architecture for 3D Atomistic Systems
Atom-JEPA is introduced, a self-supervised pretraining framework that learns latent representations from unlabeled 3D structures through complementary atom-level and substructure-level objectives inspired by joint-embedding predictive architectures that achieves state-of-the-art performance on molecular ADMET and quant...
Kasper Helverskov Petersen, Rasmus Tirsgaard, François Cornet et al.
· 1 citation