Prophet-OAME-MBD
Model overview
Interatomic potential: Energy with gradient-based forces and stress.
CPS 0.912 combines discovery, geometry optimization, and phonons. Inspect individual task results below.
Licenses are as declared by the submitter. Training dataset terms are listed separately below.
Model Info
Specifications
- Architecture
- gnn
- Outputs
- Energy with gradient-based forces and stress EFSG
- Openness
- Open source, closed data OSCD
Discovery protocol
- Training
- structure to energy, force, stress S2EFS
- Evaluation
- initial structure to relaxed energy with structure relaxation IS2RE-SR
Training Set
MPtrj: 1.58M structures from 146k materials · MIT
OMat24: 101M structures from 3.23M materials · CC-BY-4.0
Subsampled Alexandria: 10.4M structures from 3.23M materials · CC-BY-4.0
ELEMENTA: 211M structures from 9.78M materials · CC-BY-NC-4.0
Run this model
Use the submitted environment with the upstream instructions.
From a repository checkout, run uv run models/run_kappa.py --list-models to check checkpoint requirements. If this model requires one, add --checkpoint /path/to/checkpoint below. Then execute the printed command
for a phonon smoke test in the model's isolated environment.
uv run models/run_kappa.py --model prophet-oame-mbd --print-cmd --dry-runSelect an available task above to inspect this model's results.
Model Authors
Trained By
description
Prophet is a family of atomistic foundation models for materials simulation and discovery, trained on large-scale first-principles data to model energies, forces and stresses across diverse chemical systems, with extensions that explicitly incorporate spin as a fundamental physical degree of freedom. Prophet-OAME-MBD is a 62.3M-parameter model pretrained on Omat24 and ELEMENTA, including the ELEMENTA Vib expansion, and aligned on MPtrj and sAlex to the Materials Project energy reference used by Matbench Discovery. Structures and trajectories in ELEMENTA overlapping with the WBM test set were removed during post-processing.