How do i modify code related to split candidate

(I got help from a google translator. If there is any expression that looks rude, I apologize in advance.)

first of all, thank you for making great library!

I have one question.

i’m researching about gradient boosting algorithm with descretization/binning.

so, I want xgboost to use only few split candidates what I manually set.

I thought there would be code that computes weights for unique data in a variable. but it was hard to find.

here is my step.

I assumed that there will be a process of calculating leaf value for the data when building the tree. Could you please give me a hint where the code for this is located?