DPA4: A Generalist Foundation Model for the Era of On-Demand Interatomic Potentials

Tiancheng Li, Duo Zhang, Anyang Peng

Submission Type: I want my submission to be considered for poster only

Keywords: AI for science, machine-learning interatomic potentials, equivariant neural networks, molecular dynamics, materials discovery

TL;DR: DPA4 sets a new accuracy-efficiency frontier for equivariant interatomic potentials, achieving state-of-the-art results on materials and molecular benchmarks with native short-range physics and compiled conservative-force training.

Confirmation Of Submission Requirements: I submit an abstract. It uses the template provided on the submission page and is no longer than 2 pages.

Submission Number: 452