Photometric redshift prediction performance improvement (as defined in Footnote 4) with respect to the metrics with the initial training set size (0.5% of the total number of galaxies), as a function of the training set size, evaluated on a sample of 1 000 000 test galaxies. From left to right we plot: the RMSE, OLF, and NMAD for different redshift intervals, as indicated in the legend.