Suppose that you have an input volume of $32$ x $32$ x $3$ and

you have a convolution layer with $3$ x $3$ kernels with zero

padding, stride of $1,$ and $2$ filters. How many weights and

biases does the layer have?