Best split-factor for a given number of Samples? #16
Description
Is there a rule of thumb for estimating the split factor for a large number of samples? For example, what would be a good split-factor for 100k or 200k samples and the available GPU memory? What split-factor was used for the gnomAD relationship inference?