Bug for numpy backend sum method

[refraction-ray]

See source in https://github.com/google/TensorNetwork/blob/master/tensornetwork/backends/numpy/numpy_backend.py#L607.

  def sum(self,
          tensor: Tensor,
          axis: Optional[Sequence[int]] = None,
          keepdims: bool = False) -> Tensor:
    return np.sum(tensor, axis=tuple(axis), keepdims=keepdims)

Therefore, sum(tensor) raises error for tuple(None)

[mganahl]

thanks! Could you add a few lines regarding expected behaviour?

[refraction-ray]

thanks! Could you add a few lines regarding expected behaviour?

I guess return np.sum(tensor, axis=axis, keepdims=keepdims) is enough? Actually no idea why there is a tuple

[breakTheLimitsKP]

Hey, I would like to work on this, can you please assign this one to me.