Comparison of the PIT distribution before and after applying split conformal prediction on evaluated galaxies from the GalaxiesML and TransferZ samples. The evaluations are performed using the BNN models developed in this work. The PIT distribution is a qualitative diagnostic of accurate pdfs over an ensemble of galaxies. Accurate pdfs are indicated by a uniform PIT distribution U[0,1]. Our application of split conformal prediction ensures the pdfs are reliable. Before split conformal prediction, the PIT distributions are center-peaked, an indication of overdispersed pdfs; after split conformal prediction, the distributions are closer to uniform.