Delta_Kernel bug

[wahyutirta]

issue raises at this code in conv backward method
could not broadcast input array from shape (6,3,3) into shape (1,3,3)

self.delta_K[kernel_num,:,:,:] += delta[img, kernel_num, h, w] * X_padded[img, :, hstride:hstride+K_H, wstride:wstride+K_W]
inside delta kernel iteration, self.delta_kernel have 1 3x3 metric when X_padded have 6 3x3 metric.. i think it is missing np.sum before adding mutiply value to self.delta_K