Partitioned models with MLFlow

Options
dadbuz
dadbuz Dataiku DSS Core Designer, Dataiku DSS Adv Designer, Registered, Dataiku DSS Developer Posts: 16 ✭✭✭✭

Hi,

Is there a way to create and import a partitioned model through the MLFlow API?

Thank you

Tagged:

Answers

  • CoreyS
    CoreyS Dataiker Alumni, Dataiku DSS Core Designer, Dataiku DSS Core Concepts, Registered Posts: 1,150 ✭✭✭✭✭✭✭✭✭
    Options

    Hi @dadbuz
    while you wait for a more complete response, here are some resources you may find helpful:

    Please note that MLflow import is considered experimental. I hope this helps!

  • CoreyS
    CoreyS Dataiker Alumni, Dataiku DSS Core Designer, Dataiku DSS Core Concepts, Registered Posts: 1,150 ✭✭✭✭✭✭✭✭✭
    Options

    I can confirm that partitioned models are not supported through MLflow.

  • importthepandas
    importthepandas Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS Core Concepts, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 115 Neuron
    Options

    @CoreyS
    I know MLFlow integration and some of the great MLOps features are relatively new and experimental. This has been fantastic for our data scientists to extend the platform while integrating with a lot of the nice visual and MLOps features dataiku has.

    However, we love partitioned models and use them in a few production projects. We very much would love to see more support for partitioned models w/MLFlow and also for evaluations and comparisons. I think in 11 evals support partitioned models now. I'm assuming Dataiku has this on their roadmap, but if you need me to log a product idea LMK. Always glad to!

Setup Info
    Tags
      Help me…