From my view, the relay.expr.Function.params contains a list of sizes of network layers, and I can easily read them. However, the network definition (such as layer type/definition, I guess) in the expr.Function.body is confusing, it seems like a nested definition, just like expr.Function.body.args[0].args[0]… How should I extract information of each layer correctly from the ‘func’ imported from keras frontends (like resnet,…)?