Files¶

The Files tab in Registry is a centralized place to store, share, and search files of any type, including structured and unstructured data, documents, and binaries. It's commonly used to hold the data behind unstructured knowledge workflows, but you can use it for any files you need to manage and reuse.

See also:

• Supported dataset types for vector databases.
• Supported connections for unstructured data.

When unstructured data is added as part of a workflow within DataRobot, after registration is complete, the file is accessible from this page (even if the associated Use Case is deleted).

From here, you can:

For each registered file, the following columns are displayed in the table:

Select one or more files to activate additional file actions at the top of the page.

A counter displays the number of items actively selected and to the right, you have the option to:

• Link to Use Case: Link the files to one or more Use Cases.
• Share: Share the files with a user, group, and/or organization.
• Tag: Add tags to the files to use for filtering.
• Delete: Delete the files from Registry. Doing so removes the files from and Use Cases they're associated with.

View files¶

Click on a file row to open the file details pane, providing additional information about the specific item, including a preview and version history. The center panel displays a tabular preview, as well as file actions:

File metadata¶

In the right panel, the Information dropdown displays file metadata:

Schedule file refreshes¶

Scheduling allows you to automatically refresh a file on a specified schedule, so the latest version in Registry remains current with the source. Each time a scheduled refresh runs, DataRobot re-reads the file from its original source and adds it as a new version—all previous versions are listed under File versions, allowing you to compare what changed or use a previous version.

To schedule a file refresh:

1. Expand Scheduling in the right panel and click + Schedule file refresh.

2. To set a date for scheduled refreshes to begin, click the field below Start date and use the calendar picker to choose a date. The start date and cadence default to your current time.

   

3. Under Cadence, select how often you want file refreshes to occur—daily, weekly, monthly, or annually.

4. To finish creating the schedule, click Save.

Access and view the status of all scheduled refreshes by expanding Scheduling.

Open the Actions menu next to a refresh to:

• Pause/Resume: Pause or resume the scheduled refresh.
• Edit: Edit the schedule of the refresh.
• Delete: Delete the scheduled refresh.

All scheduled file refreshes appear with one of the following status badges:

Troubleshooting¶

If DataRobot cannot reach the source to perform a refresh, for example, if credentials expired or access was revoked, the schedule is disabled rather than retrying because reauthorization requires users to sign in. When this happens, you cannot activate the same scheduled refresh. Instead, you must reconnect to the source and create the schedule again.

File versioning¶

In the right panel, the Versions dropdown displays a version history of the file, including when the version was added and by whom.

Each file entry supports multiple versions, allowing you to upload a new version of the file without creating a separate entry, preserving a history of how the file changes. Click to view a specific version of the file.

Next steps¶

Once files are registered, use them as source data for other workflows or manage them alongside your other registered assets.

• Create a vector database: Use a dataset from the File Registry as the knowledge source when building a vector database for RAG.
• Create custom models: Assemble a custom model in the workshop using a collection of registered files.
• Data: Manage structured datasets in the companion Data Registry, which converts uploaded files into CSV format for use in experiments.