Skip to navigation
Explanation Lab

Use explanations

Retain the context needed to interpret and reuse explanations.
View as Markdown

Use explanations in documentation and decisions

For a reusable explanation, capture enough context for another reader to understand what was analyzed.

  • Record the deployed model and the explanation view.

  • For local explanations and graphs, retain the input values.

  • For global and symbolic explanations, identify the module and source rules.

  • For sampled LLM explanations, retain the sample size and the data context.

  • Keep screenshots of the relevant result or rule alongside the written interpretation.

  • Review generated text before including it in external documentation.

Use these workflows to query an individual record, inspect feature and module contributions, browse learned rules, trace the displayed dependency graph and review generated explanations before reuse.

See Explanation concepts for the underlying terminology and Explanation workflows for ways to use these components.