It is known that two back-to-back 3xx3 convolutional layers (a stack of two layers with no spatial pooling layers in between) have a 5xx5 effective receptive field. Assuming that both the input and the output of such a stack have 10 channels, compute the factor by which the number of weight parameters reduces if a 3xx3 stack is used instead of a single 5xx5 layer in a CNN.