Metric aggregations and Overall aggregate score

Thanks for your great work!

I’d like to check two small things:

1. In Appendix F.2. "We aggregate utility score and memorization score and use their harmonic mean for tuning the models." Does the released code already include direct computation for **Memorization**, **Privacy**, and **Utility** metrics?

2. In Table 3, how is the Overall aggregate score calculated?



