GRN in paper uses x_i / sum x_j whereas code uses x_i / mean x_j

[sebLopezCot]

Is this expected / is there some mathematical trick here or am I missing something? I am not following why in the paper there is discussion of using just a weighted sum of feature norms when it appears in the code a mean it taken in the denominator.

[SnifferCaptain]

gamma will help