In step 1, we created a variable of type symbol. We used this variable to configure the network. The mx.symbol.Variable("data") function uses data to represent the input data, that is, the input layer. We added the hidden layers using the mx.symbol.FullyConnected function; its parameters are data with the type as symbol, the name of the layer, and the number of neurons in the layer. We applied activation layers using the mx.symbol.Activation() function. At the end of the network, we added a regression output layer. In step 2, we selected the device to train the network. We can also use the mx.gpu() function for training on GPUs. 

In the next step, we trained the model. In step 4, we visualized our network. In the last step, we evaluated the model's performance.