Can it be used in regression?

[celinecp]

How can we adjust it to a regression problem?

[electricmomo]

I am also interested intot he same question

[eladmw]

Probably, just change the layers and the booster

[eladmw]

Looking at the source code more in detail, I think that you just need to choose a different loss for torch and specify a different objective function for XGBoost. "objective" = "..." --> "objective" = "..."

[eladmw]

Would you guys be happy if I rewrote the repo in LGBM, for regression and classification?