Artifacts

Artifacts are the outputs produced by pipeline steps: model files, datasets, evaluation metrics, confusion matrices, and any other files or values that a step writes and registers. BullSequana AI stores artifact metadata so you can trace exactly which run produced a given output and compare outputs across runs.

Browse artifacts

1. Select Pipelines in the left sidebar, then select the Artifacts tab.

The Artifacts list shows every artifact registered across all runs, along with:

• Name: the artifact name as defined in the pipeline.
• Type: the artifact type (for example, Model, Dataset, or Metrics).
• URI: the storage location of the artifact.
• Created: when the artifact was registered.

View artifact details

Select an artifact in the list to open its detail page. The detail page shows:

• Properties: metadata key-value pairs attached to the artifact.
• Custom properties: additional properties set by the pipeline step.
• Linked executions: the pipeline steps (executions) that produced or consumed this artifact.

Use the Linked executions section to trace an artifact back to the specific run and step that created it.

Next steps

• Browse executions to see how artifacts flow between pipeline steps.
• View run outputs to see artifacts in the context of a specific run.