A comparison of the performance of NNs using metrics averaged over the redshift range 0.3 < z < 1.5 (see Section 3.3.1). The gray shaded area corresponds to the LSST science goals. The NN-TL and NN-Combo models show comparable performance on both GalaxiesML and TransferZ, with NN-Combo meeting one more requirement than NN-TL. NN-2 performs comparably to NN-TL and NN-Combo, but with a higher outlier rate O on GalaxiesML.