Skip to content

MVP: Build evaluation harness for human-label comparison #2

Description

@alp-topcu

Goal

Measure baseline scorer behavior against reviewer labels.

Acceptance criteria

  • Human label fixture format is documented.
  • Evaluation report includes agreement metrics and per-dimension errors.
  • The harness can run locally without external model credentials.
  • Results clearly mark the baseline as unvalidated.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions