HannesStark / 3DInfomax

Making self-supervised learning work on molecules by using their 3D geometry to pre-train GNNs. Implemented in DGL and Pytorch Geometric.

HannesStark/3DInfomax Issues

RuntimeError: Error(s) in loading state_dict for PNA: size mismatch for node_gnn.atom_encoder.atom_embedding_list.1.weight: copying a param with shape torch.Size([4, 200]) from checkpoint, the shape in current model is torch.Size([5, 200]).
Closed 6 months ago
pretrain loss with negative value
Closed 6 months ago4
Help
Closed 6 months ago2
Embedding views in NTXentMultiplePositiveLoss
Closed 6 months ago1
"conda env create" problem
Closed 6 months ago1
classification yml file
Closed 6 months ago
DglPCQM4MDataset ImportError for inference
Updated 9 months ago5
Some questions about 3DInfomax
Updated a year ago1
Fine-tuning a model with moltox21 dataset - error
Updated a year ago1
Pretrained 3d model
Closed 2 years ago1
Where is the "bbbpscaffold123.pkl"? And what is the "bbbpscaffold123.pkl"?
Closed 2 years ago1
What is the MAE of QM9? Which parameter in the following operation result is Mean Absolute Error? Thank you.
Closed 2 years ago1
What model is gin.py?
Closed 2 years ago4
Fine-tuning a model with `BACEGeomol` dataset - error when collating
Closed 2 years ago4
ResolvePackageNotFound in Windows
Closed 3 years ago2
Linked video freezes
Closed 2 years ago1
Having trouble pre-training with example code
Closed 3 years ago4
generating a fingerprint
Closed 3 years ago6
having trouble training for GEOM-Mol + trained models
Closed 3 years ago5