[...]
for epoch in 1:100
[...]
@info (; epoch, train_loss=loss(train_loader), test_loss=loss(test_loader))
end
There is no function signature for loss(dl::DataLoader). The fix is obviously simple: add the required model arg--PR inbound 😉
I could also make a PR to implement CI testing on the docs examples, if that would be welcome. I like doing that for my packages, since this kind of thing has a way of popping up.
The example code at https://carlolucibello.github.io/GraphNeuralNetworks.jl/stable/ has a small problem:
There is no function signature for
loss(dl::DataLoader). The fix is obviously simple: add the requiredmodelarg--PR inbound 😉I could also make a PR to implement CI testing on the docs examples, if that would be welcome. I like doing that for my packages, since this kind of thing has a way of popping up.