That means that in inference you are reporting the F1 score, correct?

For classification it's F1 score but for other task it maybe and I don't think that's problem. we just have to log it right?

Are you using passing output_uri to Task.init?

No

are you storing the model as artifact?

Yes