label leakage
zhandand opened this issue · comments
zhandand commented
PyGCL/examples/GRACE_SupCon.py
Line 75 in 3e6357b
In the supervised contrastive learning, the nodes with the same labels are collected as the postive pairs. So in my opinion, in the constrative pretraining stage, we can't utilize all the data, otherwise the labels are leaked in the tuning stage