How to Update a Saved Trained Model Version in Azure ML RRS feed

  • General discussion

  • These instructions describe how to update a previously trained, saved model to a new version.

    Step 1: Open a scoring experiment where the saved trained model is used.

    Step 2: Select the saved trained module on experiment canvas, and click “Training experiment” link in Properties pane. This will open the latest version of training experiment. 

    To use different version of training experiment, use “View Run History” to navigate to the desired version of training experiment. Then use “Save As” to save it as new experiment.

    Step 3: Accept module upgrade for training experiment.

    Step 4: Run the training experiment.

    Step 5: Save the trained model. Check the “This is the new version of an existing trained model” box.

    Now you can use the trained model in scoring experiments, and publish these scoring experiments as web services. Note that  existing web services are unaffected until you re-publish them.

    Tuesday, December 9, 2014 12:40 AM