• num_params = connections between layers + biases in every layer = (i×h + h×o) + (h+o) = (3×5 + 5×2) + (5+2) = 32

i = 3 h = 5 o = 2

Counting no. of parameters in deep learning models