how to build the 150k streamvln dataset based on the scalevln

Thanks for your excellent work! But i have some questions.
As the title says, i can't find the code for building the 150k streamvln dataset, scalevln is a discrete dataset, how can it be transformed into vln-ce format?
