[Lightgbm] How to save models while training

In python api I can write a callback function to save models while trianing. Is there a way to achieve this goal for spark version? And I noticed that there's no eval and trianing loss printed out. Is there any params to control it? Thanks!

Start by comparing the Python callback approach mentioned in the issue with the Spark LightGBM API. Investigate whether model saving during training and evaluation/training-loss output are supported, and locate any relevant parameters or API entry points. Done should be defined as a confirmed supported approach or a clearly scoped missing capability.

Written by the indexing model from the issue text.