Assets Page
The top-level Assets page is where you browse assets across runs, not just inside a single run detail view.
Use it when your question is about the whole storage root, for example:
- which paths are generating the most assets?
- which runs have archived code, datasets, or outputs?
- which asset is reused by multiple runs?
- which archived asset should I preview or download directly?
What the page contains
The page has two main tabs:
- Overview
- Repository
At the top, it also shows storage-level stats such as:
- total runs
- runs with assets
- total asset count
- archived asset count
- archive size
Overview tab
The Overview tab groups asset usage by run path.
This is useful when you want to see:
- which paths have the most runs with assets
- how many archived assets each path contains
- the rough distribution of code/config/dataset/pretrained/output assets by path
Think of it as an operational summary rather than a file browser.
Repository tab
The Repository tab is the detailed asset catalog.
Here you can filter and inspect assets by:
- kind
- path
- archived-only vs all
- related-only vs all
- search text
- sort mode
The table exposes information such as:
- asset kind
- display name
- whether the asset is archived or only referenced
- last used time
- description and context
- source type
- source URI
- archive path
- related run count
This is the best place to answer questions like “where is that snapshot?” or “which runs reuse this artifact?”.
Asset detail page
From the Repository tab, you can open a dedicated asset detail page.
That page lets users inspect:
- identity and kind
- related paths
- description and context
- source and archive locations
- fingerprint
- preview content when supported
- which runs use the asset
How this differs from the run detail Assets tab
The run detail Assets tab answers:
- what is attached to this one run?
The top-level Assets page answers:
- what assets exist across the whole storage root?
- how are they distributed across paths?
- which asset should I open directly without first opening a run?
So they are related, but they are not the same UI surface.
Best use cases
- auditing code snapshots across many runs
- finding archived datasets or outputs without opening runs one by one
- checking which assets are shared across multiple runs
- jumping directly to preview and download flows