Closed aurorarossi closed 11 months ago
We should make the TGCN compatible with GNNChain. This requires overloading _applylayer to do what is done for GNNLayer: https://github.com/CarloLucibello/GraphNeuralNetworks.jl/blob/1a4c62ba8351cf0744bdaac3ab7821ace25fa8de/src/layers/basic.jl#L158 https://github.com/CarloLucibello/GraphNeuralNetworks.jl/blob/1a4c62ba8351cf0744bdaac3ab7821ace25fa8de/src/layers/basic.jl#L162
TGCN
GNNChain
_applylayer
GNNLayer
So this means defining
_applylayer(l::Recur{TGCNCell}, g::GNNGraph, x) = l(g, x) _applylayer(l::Recur{TGCNCell}, g::GNNGraph) = l(g) (l::Recur{TGCNCell})(g::GNNGraph) = GNNGraph(g, ndata = l(g, node_features(g)))
We should make the
TGCN
compatible withGNNChain
. This requires overloading_applylayer
to do what is done forGNNLayer
: https://github.com/CarloLucibello/GraphNeuralNetworks.jl/blob/1a4c62ba8351cf0744bdaac3ab7821ace25fa8de/src/layers/basic.jl#L158 https://github.com/CarloLucibello/GraphNeuralNetworks.jl/blob/1a4c62ba8351cf0744bdaac3ab7821ace25fa8de/src/layers/basic.jl#L162So this means defining