Open
Description
Currently, the evaluation request (evalMetrics array) in the Evaluation tab sends only one metric ('tool_trajectory_avg_score') in the request when running an evaluation. Ideally, the request should include all three relevant metrics, allowing for a more comprehensive evaluation process.
Proposed solution:
- Update the code to ensure the evalMetrics array is populated with all three metrics before submitting the evaluation request.
- Consider allowing the user to select which metrics to include, or include all by default.
Impact:
- More robust evaluation capability.
- Better alignment with expected evaluation standards.
References:
- See
evalMetricsusage insrc/app/components/eval-tab/eval-tab.component.ts.
Metadata
Metadata
Assignees
Labels
No labels