Same as Fig. 22 but for the ANN output parameters (stored in the supplementary DR3 astrophysical parameters table). In this case, we imposed that the first eight quality flags in Sect. 8 be equal to zero. The calibrations proposed in Sect. 9.2 were applied. Although a larger dispersion is observed with respect to MatisseGauguin, a general agreement exists, supporting the coherence of the two methodologically independent analysis.