Catalog of atmospheric parameters and chemical abundances from Gaia RVS spectra with convolutional neural network

These labels were derived using a convolutional neural network applied to the full Gaia data product in the form of RVS and XP spectra, parallaxes, and stellar magnitudes. As training labels, we used a set of 45000 stars with high-quality labels from APOGEE DR17.

We provide a </tt>boundary_flag</tt> that should be used to filter the stars outside of the training sample limits. We recommend using <tt>flag_boundary = '00000000'</tt> for science applications. The full table can be downloaded via following links in FITS or CSV format: <ul> <li><a href="gaia.aip.de/files/cnn_rvs_catalog/cnn_gaia_rvs_catalog_guiglion_et_al_2023_v2.fits">cnn_gaia_rvs_catalog_guiglion_et_al_2023_v2.fits</a></li> <li><a href="gaia.aip.de/files/cnn_rvs_catalog/cnn_gaia_rvs_catalog_guiglion_et_al_2023_v2.csv">cnn_gaia_rvs_catalog_guiglion_et_al_2023_v2.csv</a></li> </ul>