What are Evaluation Groups?
Evaluation groups are collections of related evaluations that can be executed together as a single unit. Instead of running evaluations individually, you can create groups that contain multiple evaluations with shared configurations, making it easier to assess your AI models across multiple dimensions simultaneously.Benefits of Using Evaluation Groups
- Batch Execution: Run multiple evaluations at once instead of adding them one by one
- Consistent Configuration: Apply the same dataset mapping and settings across all evaluations in a group
- Improved Efficiency: Reduce manual effort and time required for complex evaluation workflows
- Better Organization: Group related evaluations logically for easier management
- Reusability: Save and reuse evaluation groups for future testing scenarios
You can mix built-in and custom evaluations within the same group to create comprehensive assessment workflows.
Best Practices
- Logical Grouping: Group evaluations that assess related aspects of your AI system
- Consistent Naming: Use clear, descriptive names for groups and evaluations
- Documentation: Maintain clear descriptions of group purposes and configurations
You can use eval groups across the platform like in dataset, prompt workbench, simulation, etc.