$2.$ Now, consider a simple nnet model for the above image classification problem. $(5$ points$)$

There is $1$ hidden layer with $128$ neurons.

What are the numbers of parameters $($Weights and Biases$)$ in each layer $($input$-$hidden$-$output$)?$

What is the total number of the parameters in the model?

While we are going to use ReLU activation for the hidden layer, what do you suggest for the

output layer?