AI Models Usage
About the Data
Data Refresh Frequency: Updated daily at 8:00 am EST (UTC 1:00 pm)
Displays Data for: Last 90 days. For new installations, data will be visible after a week of scanning with version 184.
Minimum Requirements: Applications containing AI models must be scanned at least once after upgrade to version 184.
Overview
The AI Models Usage dashboard enables AI Governance in your organization. The dashboard provides information on the open-source AI/ML models that are detected during the application scans performed by Lifecycle.
By detecting the open-source AI models being used in your applications, you can create Lifecycle policies and use a structured approach to mitigate potential security and compliance risks associated with the open-source AI models at an enterprise level.
The AI Models Usage dashboard identifies the applications using AI models, provides key metadata for the model that includes model lineage, parent repository name, model name, commit hash, model file extension, type or format of the model and publishing date on the hosting platform.
Note
The AI Models Usage dashboard currently supports open-source AI models hosted on the Hugging Face platform.
Get to Know Your AI Models Usage Dashboard
The interactive dashboard provides multiple filter options to view the open-source AI model usage across your organization. You can filter on scan date, organization name, application name, application category, a specific DevSecOps stage and component type (format/ecosystem).
Stages available for filtering are release, stage-release, build, operate, compliance and source.
Models Detected
The Models Detected table displays the names of the open-source AI models found in your applications based on the filter selections. The model name (root model) displayed here is a concatenate of {{ repository }}: {{model_name}}: {{commit_hash}}: {{component_type}}: {{extension}}
Click on the model name in this table to view its corresponding metadata in the Identical Model table.
Identical Model
The Identical Model table contains metadata for the model selected in the Models Detected table. It also includes metadata for other models that are identical to the root model.
If no model is selected, it displays metadata for all models based on the filter selection.
Table column | Description |
|---|---|
Relationship | Possible values: root - indicates the model is the original model identical - indicates the identical copy of the model |
Repo | The full Hugging Face repository name in owner/name format |
Model | Name of the model |
Commit | Commit hash of the model |
Extension | The filename extension of the model |
Component Type | The format of the model, for e.g. pytorch, safetensors. |
Date | The time since the model exists on the Hugging Face repository |
Applications
This table contains the names of the applications that contain the AI model and the corresponding model name.
Table column | Description |
|---|---|
Application | The name of the application in Lifecycle containing the AI model |
Model | Model name |
Troubleshooting
Problem
Clicking on the browser Refresh button may give you the following error:
Solution
Click the Back button on your browser, from the page where you see this error, to back to the Landing page Data Insights. Select the dashboard you want to view, to reload the visualizations.
To refresh the page, click on the refresh icon on the top right, instead of the Refresh button on your browser.
Problem
No data visible on the dashboard.
Solution
Contact support. Use the cluster_id provided at the bottom of the dashboard, if available.
Problem
Other issues with the dashboard.
Solution
Contact support. Use the cluster_id provided at the bottom of the dashboard, if available.
