Graph Neural Network - Readout Method Tensorflow / Pytorch

Question

Thank you very much for reading my question. Im quite new to NN and have recently got to know about graph neural network and am trying to learn about graph classification(regression). Im using dgl library, and is it good for beginner ? the part that puzzles me most is after you conv the nodes and readout the data to be inputted to an ordinary feed-forward layer. At the moment, I am just using as the simplest as listed :

out_ = dgl.mean_nodes(g, 'x') 
out_ = Dense_Layer(out_)

does many readout method : [min, max, mean, sum, ...?] just group all the nodes into one value ? I am looking for some method to readout the nodes to some larger dimension whilst m struggling to even know what to search on google, is this topic about pooling also... would very appreciate if you may point out some direction/ study/ some code to look at that I can look into.

Thank you very much.