The current implementation of file export/import for sparse tensors has some inefficiencies, as it writes/reads one line at a time. To address this, I propose using numpy.savetxt and numpy.loadtxt to provide considerable speedup.

sparse_export.py: Compares the current export functionality to a new option using numpy.savetxt.
sparse_import.py: Compares the current import functionality to a new option using numpy.loadtxt.
I will create a merge request with the proposed changes once this issue is acknowledged.