Closed reshinthadithyan closed 4 years ago
In case no node features are given, you have multiple options:
transforms.OneHotDegree()
torch.nn.Embedding
.
The first option is a good choice in inductive training scenarios, while the latter two are good choices in transductive/semi-supervised learning tasks.Thank you, @rusty1s .
❓ Questions & Help
Say I have a Graph, with Nodes which doesn't hold individual Node Features. How can I pass it through a GCNs?