I tried many things such as np.asscaler and squeeze and a lot but it was in vain. any hint, please
You already wrote a subroutine to handle that case, right? It was called conv_single_step. Did that code pass the test cases in the notebook?
Of course what that error message is telling you is that the RHS of that assignment statement is an array, not a scalar value. If you look at what happens in conv_single_step, there needs to be a sum following the elementwise multiply.