Pretrain using a new dataset

[kdmsit]

Hey, I want to use this pretraining strategy to pre-train the GNN model on a new chemical dataset from scratch. What I have understood from the code, all the datasets are preprocessed in x, edge_att,edge_index, id,y, and fold. Could you let me know how did you preprocess it and share the code so that I can preprocess my dataset and pre-train a GNN model from scratch?

[engmubarak48]

I also have a similar question. I downloaded and checked their data and it seems they preprocess someway different from that of OGB benchmark dataset. Wanted to ask if we can pre-train the models on the datasets from the OGB benchmark by using the PyTorch geometric loaders.

[David-Ttao]

OMG, I also the same question. do u have solved it?