I’ve set up Tuner with a CometLoggerCallback. However, when I call train.report(metrics) which is identical to train.report(metrics, checkpoint=None) comet will reports many COMET WARNING: The given value of the metric checkpoint_dir_name was None; ignoring for every report call.
I do not want to keep many checkpoints and actually pass a value there. I tried to change the log level of the comet_ml logger to ERROR, but it did not stop the warning message - and I actually do not want to modify comets logging level above Warning - only if I have to.
Does anyone know how I can suppress this error in my trainable function?