def mse_loss_grad(y, t): loss_ = np.sum(np.subtract(t-y))*(2/y.shape[0]) print("Derivative MSE: ".format(loss_)