CatBoostLSS implementation still of interest?

[StatMixedML]

Dear Community,

Given that CatBoost does not support GPU training of custom loss functions, I want to have your opinion on whether or not a CatBoostLSS implementation would still be of interest. Let me know!

[kmedved]

Thanks for your work on this @StatMixedML. I voted yes, as Catboost currently still has the best-out-the-box results (without any tuning). However, insofar as there is a lot of work required to maintain each of CatBoostLSS, LightGBMLSS, and XGBoostLSS, it would probably make more sense to focus on one of them and make it a best-in-class library rather than dividing efforts across all three.