Open MatSci ML Toolkit is a framework for prototyping and scaling out deep learning models for materials discovery supporting widely used materials science datasets, and built on top of PyTorch Lightning, the Deep Graph Library, and PyTorch Geometric.
MIT License
144
stars
20
forks
source link
Allow `hidden_irreps` to be passed through the mace wrapper #244
There are usecases in which the hidden_irreps may need to be passed in as an o3.Irreps module, such as Irreps("16x0e+16x1o+16x2e"). This update allows the hidden_irreps kwarg to be used in the mace wrapper, and defaults to construct this from atom_embedding_dim if not supplied.
There are usecases in which the
hidden_irreps
may need to be passed in as ano3.Irreps
module, such asIrreps("16x0e+16x1o+16x2e")
. This update allows thehidden_irreps
kwarg to be used in the mace wrapper, and defaults to construct this fromatom_embedding_dim
if not supplied.