You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, thanks for releasing this comprehensive codebase combining both DeepGCNs paper and DeeperGCNs paper. I was wondering whether you have experimented with the architectural advancements from DeeperGCNs on the 3D point cloud tasks presented in DeepGCNs? Do the new generalized aggregators, message norm, etc. improve performance for 3D point cloud tasks, too?
The text was updated successfully, but these errors were encountered:
Hi @chaitjo. Thanks for your suggestion. We have not tried the proposed methods of DeeperGCN on 3D yet since for 3D point cloud tasks such as s3dis require longer training time and larger GPU memory. But it is in our plan. I would let you know once we get any results.
Hi, thanks for releasing this comprehensive codebase combining both DeepGCNs paper and DeeperGCNs paper. I was wondering whether you have experimented with the architectural advancements from DeeperGCNs on the 3D point cloud tasks presented in DeepGCNs? Do the new generalized aggregators, message norm, etc. improve performance for 3D point cloud tasks, too?
The text was updated successfully, but these errors were encountered: