LeNet neural net graph interpretation


Iam using using the LeNet neural network architecture on the MNIST databse and I came up with “MXNetNodeGraphPlot”. I do understand the network graph from Information[LeNet, #] & /@ {"SummaryGraphic"}, but how is the MXNetGraph interpreted?, why does the convolutinal layers and fully conected layers have three inputs? and why is this different from the "SummaryGraphic"?

enter image description here

Here is the full code:

LeNet = NetModel["LeNet Trained on MNIST Data", 
  "UninitializedEvaluationNet"]
Information[Net, #] & /@ {"SummaryGraphic"}
Information[Net, #] & /@ {"MXNetNodeGraphPlot"}