Skip to main content

Benchmark Evaluation Runs

Status

Owner: TBD · Status: draft · Canonical source: flezi.app.frontend.main/src/pages/EvaluationRuns.tsx

Who is this for?​

Teams validating agent upgrades prior to production release.

What can you do?​

  • Execute automated benchmark runs against golden evaluation datasets.
  • Review aggregate performance metrics and inspect individual failure cases.
  • Perform A/B testing and regression checks between agent versions.

Steps​

  1. Select Evaluation Runs and click New Evaluation Run.
  2. Select the target agent and version under test.
  3. Bind the evaluation dataset and active evaluators.
  4. Click Start Evaluation.
  5. Monitor background execution and review the synthesized score report.