Can you provide a simple example that reproduces the results of the paper or can be run?

[youngfish42]

The many task scenario involved in this project is a very interesting new scenario.
I noticed that the current algorithm does not provide a demo, and I hope to apply to the author for a simple case that can be used to reproduce the results of the paper or can be run.

In addition, I would like to ask, there seem to be two different grouping methods in dg_utils.py, get_grouping_score and get_soft_grouping_score. What is the difference between them? And which method was chosen in the paper?