Deep neural networks provide a novel and highly successful avenue to calculate notoriously challenging electronic ground state wave-functions for molecular Hamiltonians. Whereas initial architectures were trained separately for any given molecular Hamiltonian, neural networks have recently been proposed which simultaneously calculate the ground state wave-functions for multiple geometrical arrangements of a given molecule. This work constitutes a further foray in that vein, proposing a neural architecture which explicitly recognizes rotations of global rotations of a molecule's atomic positions. Such explicit recognition of spatial symmetries has a celebrated history in related fields of research. While our proposed neural wave-function is indeed able to treat spatial symmetries properly, it nevertheless fails the tests of empiricism, being far from competitive with the state-of-the-art, which we corroborate with a series of numerical experiments.