Bug in the code

in NLPS -> course 2 -> week 4 -> Assignment -> Ex.4 -> we should separate the l1 computation into l1_relu and l1 variables. l1_relu should be used for gradient computations of grad_W1 and grad_W2, while the original l1 should be used for grad_b1 and grad_b2. Look into this quicker. thanks!