I could be missing something obvious, but it looks to me as though the polycrystalline total and electronic dielectric constants, poly_total and poly_electronic, on the web interface and obtained using the API, are incorrect. The dielectric tensor matrices match those in the separately published dataset (https://datadryad.org/resource/doi:10.5061/dryad.ph81h/3), but the dielectric constants do not. E.g. for mp-567290, the electronic tensor is [7.09, 7.09, 6.94] in both cases, but poly_electronic is 7.04 in the original dataset (which seems right) but 2.35 in the MP (which doesn’t). Additionally, poly_electronic = poly_total for all the entries in the MP.