This repository provides a trained machine learning (ML) potential developed for predicting energies and forces in molecular crystals. The model enables fast and accurate geometry optimization (relaxation) of crystal structures, supporting in silico workflows for materials discovery.
Use the relaxation.ipynb file to run the relaxation on test.cif.
The dataset.ipynb shows how to extract the structure, energy, forces and stress from test.cif.
The trained model is in m3gnet folder