search

divelab / DIG

A library for graph deep learning research
https://diveintographs.readthedocs.io/
GNU General Public License v3.0
1.86k stars 281 forks source link

Optimal hyper-parameters for ComENet #155

Open yueliu1999 opened 2 years ago

yueliu1999 commented 2 years ago

Congratulations on your NeurIPS 22 paper 'ComENet: Towards Complete and Efficient Message Passing for 3D Molecular Graphs', pretty great work! Could you provide the optimal hyper-parameters for your proposed ComENet like SphereNet?

limei0307 commented 2 years ago

Hi @yueliu1999,

Thanks for your interest in our work! We are reproducing the result with our modularized APIs in DIG and will provide the trained models like SphereNet in the following weeks.

Best, Limei

yueliu1999 commented 2 years ago

Thanks!

zhangyn1415 commented 1 year ago

Hi， Is there trained models or optimal hyper-parameters for the QM9 dataset can be provided now？

mseok commented 1 year ago

Hi, I'm currently looking for the optimal hyper-parameters of ComeNet for the QM9 dataset for my work. It's hard to find all the hyper-parameters with a grid search, so can you tell me which combination you've tested? Or if it is possible, can you provide hyper-parameters?