Verta MLflow Integration
Overview
Verta Model Catalog helps you create a unified model inventory, regardless of where the model was originally built or trained. To simplify the process of importing models into Verta Model Catalog, we have developed a Verta MLflow integration that allows you to batch import MLflow models with just a few clicks.
This feature is in ALPHA meaning we would love for you to try it but caution that it may sometimes behave oddly. Please provide any feedback you have to help@verta.ai and keep an eye out for an email from our product team after using this feature.
How to use MLflow batch import of models in Verta Model Catalog
Click on the “Import Models” button in the Model Catalog homepage.
Connect to MLflow
Pick MLflow as the source for import
Provide your MLflow credentials to connect and fetch the models. The credentials required are;
MLFLOW_TRACKING_URI
DATABRICKS_HOST
DATABRICKS_TOKEN (Your API Key)
After successful authentication, your MLflow model list will be visible in the UI
Select the models you want to import
You can manually select the models you wish to import from the UI list view
You can also use advanced filtering capabilities to select a subset of models. Filtering options include model name, tag, or creation date range
Begin batch import of the selected MLflow models
Verify the model selection/filtering is as expected and then long press the import button to start the batch import job
Please note that the system will not import a model if
It has a duplicated name
It has no version
In the case that the version has an error
If the version has no artifacts, or it's not compatible with the Verta system
If Verta cannot map the model stages correctly
Track the status of batch import
After initiating the batch import job, you'll be redirected to the model import status page to track progress
In this view, you can track the import status of the models, including any errors or failures
You can access the import status page at any time from the 'Import Model' button
Review the imported models in Verta Model Catalog
Once the import is complete, you can access the models and their associated artifacts in Verta Model Catalog
The models will be automatically tagged as 'mlflow_import,' making it easy for you to search and filter them
Last updated
