there’s an example notebook at https://colab.research.google.com/drive/1ySm6HYvALerDiGmk6g3pDz68V7fAtrQH#scrollTo=thvzNpmahNNx i think
in the notebook you can see they set a temperature and perform probabilistic sampling that may reduce the performance of the model by avoiding the most likely continuations you can also see how it generates changes continuously and could likely somehow be used to discover new file names
