search

TUM-DAML / gemnet_pytorch

GemNet model in PyTorch, as proposed in "GemNet: Universal Directional Graph Neural Networks for Molecules" (NeurIPS 2021)
https://www.daml.in.tum.de/gemnet
Other
184 stars 29 forks source link

QM9 dataset #3

Closed smiles724 closed 3 years ago

smiles724 commented 3 years ago

Hi, I notice you implemented experiments in extensive datasets. However, most of your baseline models show their performance in the popular QM9 dataset. Can you please provide this sort of information so that we can have a more clear understanding of how well your model is? Thanks.

gasteigerjo commented 3 years ago

GemNet focuses on molecular dynamics and force predictions, not on other molecular properties. QM9 only contains equilibrium configurations, on which force predictions wouldn't make any sense (all forces are 0).

But please do feel free to run tests. The project structure is rather similar to DimeNet's code, so you could use some stuff from the data container there. I wouldn't expect any substantial improvements out of the box. But with some hyperparameter tuning you might achieve something.