gemnet_oc

GemNet-OC: Developing Graph Neural Networks for Large and Diverse Molecular Simulation Datasets

Johannes Gasteiger, Muhammed Shuaibi, Anuroop Sriram, Stephan Günnemann, Zachary Ulissi, C. Lawrence Zitnick, Abhishek Das

[arXiv:2204.02782]

When running inference with a pretrained GemNet-OC model, make sure that the scale_file path is correct in the config, otherwise predictions will be inaccurate.

Model	Val ID 30k Force MAE	Val ID 30k Energy MAE	Val ID 30k Force cos	Test metrics	Download
gemnet_oc_2M	0.0225	0.2299	0.6174	S2EF | IS2RE | IS2RS	config | checkpoint | scale file
gemnet_oc_all	0.0179	0.1668	0.6879	S2EF | IS2RE | IS2RS	config | checkpoint | scale file
gemnet_oc_large_all_md_energy	0.0178	0.1504	0.6906	S2EF	-
gemnet_oc_large_all_md_force	0.0164	0.1665	0.7139	S2EF | IS2RS	config | checkpoint | scale file
gemnet_oc_large_all_md_energy + gemnet_oc_large_all_md_force	-	-	-	IS2RE	-

Citing

If you use GemNet-OC in your work, please consider citing:

@article{gasteiger_gemnet_oc_2022,
  title = {{GemNet-OC: Developing Graph Neural Networks for Large and Diverse Molecular Simulation Datasets}},
  author = {Gasteiger, Johannes and Shuaibi, Muhammed and Sriram, Anuroop and G{\"u}nnemann, Stephan and Ulissi, Zachary and Zitnick, C Lawrence and Das, Abhishek},
  journal = {Transactions on Machine Learning Research (TMLR)},
  year = {2022},
}