## ---------------------------------------------forward pass----------------------------------------------

## x is  (N, D) and W1 is (D, hidden) and hence, Z1 is (N, hidden)
z1 = np.dot(x, W1) + b1
## applying activation function
a1 = relu(z1)
## Z1 is (N, hidden) and W2 is (hidden, C) and hence Z2 be should (N, C)
scores = np.dot(a1, W2) +  b2