Consider the following setup for a convolutional neural network.

Input image is $13$x$13$

Layer $1$: Convolutional layer with one $4$x$4$ filter with a stride of $1$

Layer $2$: Max pooling layer with $2$x$2$ filter and stride $2$

Layer $3$: Output dense layer with $4$ neurons.

You do not use zero$-$padding anywhere in the network. How many parameters are there in layer $3?$

Group of answer choices

$104$

Cannot say

$100$

$4$