The model list displays all registered machine learning models in the current Workspace and serves as the unified entry point for model assets. You can search and filter models here, and navigate to the model details page.
Prerequisites
Before you go to the model list and perform operations, confirm the following:
You have activated DataBuddy, and the current account has joined the target Workspace.
The current account has USE CATALOG and USE SCHEMA permissions on at least one Catalog of type Model. Otherwise, the model list displays a placeholder message without specific content.
To import external models, the account must have the CREATE MODEL permission on the target Catalog/Schema and have access to an available job cluster.
Accessing the Model List
In the DataBuddy console, click Data Science → Model Management in the left sidebar to go to the model list page.
Description of Fields in the List
Field | Description |
Name | Model name, in the format of model_name. Click the name to go to the model's details page under Catalog. |
Catalog | Name of the Catalog to which the model belongs. Click it to go to the Catalog details page. |
Schema | Name of the Schema to which the model belongs. Click it to go to the corresponding Schema details page. |
Modification time | Time of the model's most recent metadata change, including version registration and tag changes. |
Owner | Model creator, which can be changed through transfer in Catalog. |
Latest Version | Latest version number of the model |
Description | Model description, which can be modified on the details page. |
Search and Filter
The model list provides the following search capabilities:
1.Keyword Search
Enter a keyword in the search box to match model names in real time. When a match is found, the model is displayed. In the search results, the parts that match the keyword are highlighted.
2.Filter by Owner
A dropdown filter for owners is provided to the right of the search box:
The options are populated from a deduplicated list of owners extracted from models accessible to the current user. All owners are selected by default.
After you select an owner, the list displays only that owner's models. Keyword search and owner filtering take effect simultaneously.
Registering a New Model
The model list itself does not provide Create Model. Models automatically appear in the list after they are registered through Autologging, the MLflow API, or the experiment run details page UI. For details about the three registration methods, model signature requirements, and version increment rules, see Model Version Management. Exporting a Model
To export a managed model to a local or external environment, use the MLflow API in Notebook. In the DataBuddy Notebook kernel, the Registry connection is automatically configured, so you do not need to manually set the URI:
import mlflow
# The DataBuddy Notebook kernel has automatically configured the Registry, so you can download it directly.
# model_name: The fully qualified name of the model, in the format of <catalog>.<schema>.<model>, for example, catalog.schema.model.
# model_version: The model version number (integer), such as 1, 2, or 3.
mlflow.artifacts.download_artifacts(
f"models:/{model_name}/{model_version}",
)
Use Limits
The model list displays only models in the current Workspace for which the current user has USE CATALOG and USE SCHEMA permissions. Models without these permissions are not displayed.
FAQs
Q: Why can't I find the model I just registered in the model list?
A: Check the following: 1. You have the USE permission on the Catalog / Schema to which the model belongs. 2. The model registration process has fully completed. Registration is asynchronous, so refresh the list to check. 3. Whether the search is performed in the wrong Workspace.
Q: Is there a difference between deleting a model from the list and deleting it from Catalog?
A: No. A model is always an asset in Catalog within DataBuddy. The delete operation directly acts on Catalog metadata and cascades to delete all versions of the model. This operation cannot be undone.
References
Model Tags and Lifecycle