-1

It is recommended to use mlflow on .py file instead of a jupyter nootebook. Why?

If I use it directly on the notebook, will I have problems in the future?

Laura
  • 1
  • What do you mean, "It is recommended"? By whom? – Matt Hall Sep 13 '22 at 14:50
  • Here is one person recommending [that](https://stackoverflow.com/a/57245147/8508004) and in the context it is clear why as it ties up the notebook viewing the Tracking UI. However, in looking, I see several examples using notebooks to introduce use of MLFlow. I'll add that in general there is a view when you need flexibility and recombinability that python scripts, as opposed to Jupyter notebooks harboring Jupyter code, are better in the long run. See ['The case against the jupyter notebook'](https://towardsdatascience.com/the-case-against-the-jupyter-notebook-d4da17e97243) and follow-up – Wayne Sep 13 '22 at 21:21

1 Answers1

0

It’s works fine in a notebook, in fact that’s where it’s arguably the most helpful. If you use it in a notebook, you will have no problems in the future.

Mike L
  • 1
  • 1