Limits, approximation and size transferability for GNNs on sparse graphs via graphops

Open in new window