in Tensorflow 2.3, how to train multiple models with one total loss?

Question

I\'m trying to train multiple models each other. There are more than two models, and each model has individual cross-entropy loss. Also, there are kl divergence loss for these m