I think we can drop the constant_hessian_value in the SplittingContext, and always assume the constant hessians value is 1. We just have to set the gradients value accordingly, to have consistent leaf values.
This is what we do for LS loss already. WDYT @ogrisel ?
I think we can drop the
constant_hessian_value
in theSplittingContext
, and always assume the constant hessians value is1
. We just have to set the gradients value accordingly, to have consistent leaf values.This is what we do for LS loss already. WDYT @ogrisel ?