[feature] Evaluators - Debugging

[bekossy]

Description

This PR enhances the Evaluation UI for both auto and human evaluations, improving user experience and workflow efficiency.

Related Issue

Closes AGE-587

Related PR

Commons PR #108

Key Changes

• Auto-Evaluation: Redesigned interface and added filters for easier navigation.
• Human-Evaluation: Improved flow, feedback, and layout for better usability.
• Evaluator Management: Full UI overhaul, added evaluator debug/test feature.

QA Instructions

• Evaluators:

  • Create New Evaluators: Ensure new evaluators can be created without issues.
  • Test Evaluator Debug Feature: Thoroughly test the debug feature by selecting a test set and a variant, and confirm the variant runs successfully.
  • Customize Evaluator Output: Verify that the advanced settings allow for proper customization of the evaluator output.
  • Use Evaluator Filters: Apply all available filters to ensure you can accurately find the desired evaluator in the suggestions list.
  • Perform C.R.U.D. Operations: Test the creation, reading, updating, and deletion (CRUD) operations on evaluators to confirm full functionality.

• Auto-Evaluation:

  • Create New Evaluations: Confirm that new auto-evaluations can be created successfully.
  • Status Update Check: Verify that the evaluation status updates correctly and does not remain stuck at 0s.
  • Filters, Sorting, and Editing: Test all filter, sort, and edit functions to ensure they work as expected.
  • Batch Evaluation Creation: Successfully create multiple evaluations at once and confirm they are processed correctly.
  • UI Interaction: Interact with every visible element in the UI and ensure everything functions properly.

• Human Evaluations:

  • Create New Human Evaluations: Test the successful creation of new human evaluations.
  • Delete Evaluations: Ensure that both single and multiple evaluations can be deleted without errors.
  • UI Interaction: Engage with every visible UI element and confirm all components are working as intended.