Skip to content

Evaluation: Categorize scores with labels #1071

Description

@AkhileshNegi

Is your feature request related to a problem?
Eval scores are currently presented as raw 0–1 floats without categorical labels in the data. This makes it difficult for API consumers to interpret scores as low/good/high without implementing their own thresholds, which the team agreed should be included in the data payload.

Describe the solution you'd like

  • Define traffic-light buckets that map score to category label (e.g. low / good / high) for all judge metrics.
  • Include the category label in the eval results data payload.
  • Use a consistent scale (0–1) across all judge metrics.
  • Future UI updates to color-code metric cards red/orange/green with an info button for recommendations.

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Type

No type

Projects

Status
To Do

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions