I am trying to understand this library. Notably how budget scale runtime:
How does run time (measured in real time) scale with budget ?
When benchmarking it seems you go from 56x speed up to 41x speed up when you go from budget 1 to budget 2.1. This speed up look insane compared to lgbm that is considered very fast. Can you clarify how is this speed up obtained ? (it seems you compare to 100 lgbm runs)
I am trying to understand this library. Notably how budget scale runtime: