Main model. A spectrum is given as input. An encoder maps it to the latent space, divided into chemical latent space z=(z_M, z_C, z_α), and auxiliary latent space w. Three decoders map different components of z, along with w, to separate regions of the input spectrum. A linear transform is applied to the auxiliary space to obtain (T_eff, log g). The discriminator is trained to predict the nonchemical parameters (T_eff, log g) from z, while the encoder is trained to make this impossible for the discriminator to predict these parameters from z alone. The dotted gray arrows represent the absence of gradient propagation (see Sect. 2.4). At inference time the decoders, the discriminator and the linear transform module are discarded.