jorgemunozl/mace_omat_medium

MACE-OMAT medium — Foundation Model

Vanilla MACE-MP-0 OMAT medium foundation model — a universal equivariant message-passing neural network potential trained on the Materials Project database. Covers the periodic table up to heavy elements.

---

Quick start

from mace.calculators import mace_mp
from ase.io import read

calc = mace_mp(model="medium", device="cpu", default_dtype="float64")

atoms = read("your_structure.extxyz")
atoms.calc = calc

energy = atoms.get_potential_energy()   # eV
forces = atoms.get_forces()             # eV/Å

Architecture

E(3)-equivariant message-passing neural network with body-ordered Atomic Cluster Expansion (ACE) basis. 128 channels, cutoff 5 Å, 2 interaction blocks, body order 3.

Parameter	Value
Channels	128
Cutoff	5.0 Å
max_L	1
max_ell	3
Interactions	2
Body order	3

Inference-only upload

This model is the unmodified foundation checkpoint. Uploaded here for programmatic download. No fine-tuning, no changes.

Citation

@article{batatia2023foundation,
      title={A foundation model for atomistic materials chemistry},
      author={Ilyes Batatia and Philipp Benner and Yuan Chiang and Alin M. Elena
              and Fabian Zills and Gábor Csányi},
      year={2023},
      eprint={2401.00096},
      archivePrefix={arXiv},
      primaryClass={physics.chem-ph}
}

MACE-Universal by Yuan Chiang, 2023, Hugging Face, Revision e5ebd9b, DOI: 10.57967/hf/1202, URL: https://huggingface.co/cyrusyc/mace-universal