AI and the World's Hard Problems · Chemistry & materials · arXiv · 2026-10-19
AceFF: A State-of-the-Art Machine Learning Potential for Small Molecules
Machine learning interatomic potentials (MLIPs) often generalize poorly across diverse chemical spaces, which limits their use for small-molecule drug discovery. AceFF refines the TensorNet2 architecture and trains it on a comprehensive dataset of drug-like compounds. The authors report that it balances fast inference with DFT-level accuracy and is state-of-the-art for organic molecules in the speed and accuracy range relevant to drug discovery.