MLFlow Version update & python code env update

[omarh2m]

Hello Dataiku community!

I've trained a model using our Dataiku 12.2.0 version and later exported it as an MLflow to deploy it onto an AKS cluster. However, our cybersecurity team has advised updating the Python libraries and the MLflow version. As a platform admin profile user, but without backend access, how can I proceed with this update?

Thanks in advance!"

[SarinaS · Accepted solution]

Hi @omarh2m,

In general, to update the packages associated with your model, you would update the packages within the code environment used to train the model, and then retrain the model on the code environment that contains the updated packages. However, for MLFlow, there are specific limitations on the package versions that can be used. You can see a full list of these here:
https://doc.dataiku.com/dss/latest/mlops/mlflow-models/limitations.html#limitations-and-supported-ve... 

So you would want to stick with a recommended or tested set of packages. 

I hope that information is helpful. Please let me know if you have any questions about this! 

Thank you,
Sarina