explain the automatic configuration of the batch size and handling of out-of-memory errors because of too high batch sizes
maybe shortly reference optimizers and loss function sections with the hyperparameters discussed there, e.g.g since optimizer parameters are also considered as hyperparameters