MLflow auto logging callback only applies to train but not evaluation?

I was able to follow the MLflow AutoLogging feature of air.callback() to log metrics and params from mlflow. However, only the training metrics/params are logged but evaluation metrics/params were not. Is there a way to log evaluation as well?

Hi @zhh210 ,

Is this problem specific to MLflow integration or do you see it in other loggers (e.g. tensorboard / Wandb as well?

Also, can you provide a small script to repro the problem?