Examine the models to determine whether it was necessary to avoid using primitives from topomodelx.base.conv. If it turns out that avoiding them was not justified, update the models by incorporating these topomodelx.base.conv primitives.
[ ] UniGCNIILayer
[ ] UniGINLayer
[ ] UniSAGELayer
[ ] HyperGAT (Note: HyperGat precomputes normalization during model initialization. Hence, it is impossible to use for graph-level tasks when the incidence matrices are different.)
[ ] HNHN (Note: HNHN takes as input incidence matrix to employ normalization possibly, it can be moved into forward pass to allow different incidence matrices)
[ ] HyperSAGE (Note: 1) not train/validation modes as well as node sampling, 2) slow due to loops and .to(device) to allow gpu )
Examine the models to determine whether it was necessary to avoid using primitives from topomodelx.base.conv. If it turns out that avoiding them was not justified, update the models by incorporating these topomodelx.base.conv primitives.