Isayev Lab at CMU

We develop machine learning methods for chemistry and materials science at Carnegie Mellon University. Our research spans neural network potentials, molecular generation, and AI-driven discovery of molecules and materials.

PI: Olexandr Isayev

AIMNet2 Models

AIMNet2 is our neural network interatomic potential for fast, accurate molecular simulations. It predicts energies, forces, charges, and Hessians at near-DFT accuracy, orders of magnitude faster.

Model Description Elements
aimnet2-wb97m-d3 General purpose (wB97M-D3) 14: H,B,C,N,O,F,Si,P,S,Cl,As,Se,Br,I
aimnet2-2025 Improved intermolecular (B97-3c) 14: H,B,C,N,O,F,Si,P,S,Cl,As,Se,Br,I
aimnet2-nse Open-shell / radicals reactive (wB97M) 14: H,B,C,N,O,F,Si,P,S,Cl,As,Se,Br,I
aimnet2-pd Palladium chemistry (wB97M-D3) 14: H,B,C,N,O,F,Si,P,S,Cl,Se,Br,Pd,I
aimnet2-rxn Reaction modeling, TS, NEB (ωB97M/def2-TZVPP) 4: H,C,N,O

All models include 4 ensemble members in safetensors format for uncertainty estimation.

Quick Start 

pip install "aimnet[hf]"

from aimnet.calculators import AIMNet2Calculator

calc = AIMNet2Calculator("isayevlab/aimnet2-wb97m-d3")
results = calc(
    {"coord": coords, "numbers": atomic_numbers, "charge": 0.0},
    forces=True,
)

Links

Citation 

@article{anstine2025aimnet2,
  title={AIMNet2: A Neural Network Potential to Meet your Neutral, Charged, Organic, and Elemental-Organic Needs},
  author={Anstine, Dylan and Zubatyuk, Roman and Isayev, Olexandr},
  journal={Chemical Science},
  year={2025},
  publisher={Royal Society of Chemistry},
  doi={10.1039/D4SC08572H}
}