LSTM Aggregator in GCN

[CodeGeek04]

I want to add an LSTM Aggregator into my GCN architecture. Number of input features in 384. Can someone please help me do so, preferably by providing relevant code

[LukeLIN-web]

Use GCNConv(aggr = LSTM) ?
https://github.com/pyg-team/pytorch_geometric/blob/master/torch_geometric/nn/conv/gcn_conv.py#L243

[rusty1s]

Yeah, this is the way to go. You should be able to customize the aggregation of GNN layers by using any Aggregation module listed here.