Learningrate 0.002 0.001 0.002 Batchsize 200 200 500 k {1,5,10,15,25,30} {1,5,10,15,25,30} {1,5,10,15,25,30}BContrastivelossderivation

In this section, we include the hyperprameter details we use in this work, e.g. We apply different sets of hyperparameters, with respect to the dataset the neural topic modelistrainedon. We demonstrate the effect of changing the number of tokens chosen for sampling. For visibility, we normalize them to one common scale before plotting them in Fig 1. It can be seen that the performance initially increases aswe select more tokens from the reconstructed output to substitute forthedrawnsample.