Exploring metadata in the run view#

When you click on a run in the runs table, the run view opens.

Here, you can:

Details view#

In the left pane, underneath the run ID and name, click Details to access the details view.

Here, you can see basic metadata about the run itself: time data, size, owner, and descriptive information.

You can also:

In the left pane, you can switch between dashboards to see the run metadata displayed in different ways.

The metadata is presented in a folder-structure, with all namespaces and fields exactly as they were logged.

You can also see the three automatically generated namespaces:

Namespace	Description
`monitoring`	Hardware consumption metrics, organized into subfolders per process stderr and stdout streams
`source_code`	The code of the executed script Git information Jupyter Notebook snapshots
"sys"	The system namespace, which contains basic metadata about the run (see Details view above)

This section visualizes all series of numerical values – like metrics and losses during training – as graphs.

You can filter the charts by typing a string into the Filter widgets input box.

If image files or figure objects were logged for the run, they appear in this section.

You can download the images from here, or inspect the evolution of an image series in the gallery view.

This section visualizes system metrics and hardware consumption logged during your run.

This section displays source code information, including:

Code snapshot: The filename and code of the script that was executed.
Git: Git information, if you have Git initialized in your project directory.
Notebook snapshot: Jupyter Notebook checkpoints, if you have the neptune–notebooks extension set up.

This section displays any files that have been logged as artifacts, with the track_files() method.

Click on the files to view the artifact metadata, such as hash, size, and path.