search

snap-stanford / ogb

Benchmark datasets, data loaders, and evaluators for graph machine learning
https://ogb.stanford.edu
MIT License
1.89k stars 397 forks source link

what does JK stands for in conv.py #307

Closed epsilon-deltta closed 2 years ago

epsilon-deltta commented 2 years ago

https://github.com/snap-stanford/ogb/blob/master/examples/graphproppred/mol/conv.py

class GNN_node(torch.nn.Module):
    def __init__(self, num_layer, emb_dim, drop_ratio = 0.5, JK = "last", residual = False, gnn_type = 'gin'):

I'm sorry to ask this stupid question. but i wonder the full name of JK. I understood how it works. lt's related to how to return the node_representation. Thanks in advance

rusty1s commented 2 years ago

It refers to the usage of Jumping Knowledge, see here.

epsilon-deltta commented 2 years ago

aha! thanks for the fast answer