Is there any way to find the optimize Number of Topics ?

[Vela-zz]

I have read your paper and found that in you Paper Figure 6., when the number of topics go from 20 to 100, there are still considerable gain in INFORMATION GAIN index.

so How many Topics are there till the INFORMATION GAIN becomes convergence?

and another question,

I found it is difficult to control the number of topics by HyperParameter because of HDBSCAN, what is the method you use to control the number of topics TOP2VEC would generate when you paint Figure 6. ? or if I change the CLUSTER MODEL to K-Means, are there any obvious shortcomings ?