Upgrade to Spark 3.5

[indranilr]

Any plans to upgrade to Spark 3.5 and the latest Xgboost 2.x ?

[jsleight]

I'm happy to review/accept PRs for either/both upgrades if anyone wants to make one 👍

[indranilr]

Probably a conflict of interest question and kindly bear with me for asking this, @jsleight would you recommend any alternative for inference optimized serialization of xgboost spark pipelines ( I see that MLFlow now also deprecating mleap support and recommending ONNX as format, although not sure how does one get to ONNX from Spark Xgboost)

[jsleight]

I've started to use onnx for some stuff alongside mleap as well. I've mostly done TF/pytorch stuff and not sure if I've done xgboost on onnx yet. onnx docs have decent guides to get started on the conversion APIs