• Review contributor evaluations of model-generated responses to ensure adherence to project-specific guidelines and rubric dimensions. • Verify that contributors consistently apply all instructions, constraints, and evaluation criteria when assessing model responses. • Confirm that contributors accurately identify factual errors, hallucinations, or missing information in model responses, ensuring evaluations reflect whether responses fulfill the user query. • Ensure contributor-selected options align appropriately with rubric dimensions and are justified by the evidence present in the model response. • Review open-text comments provided by contributors, ensuring they identify areas of opportunity, weaknesses, or limitations that may not have been captured by rubric criteria. • Provide detailed feedback on contributor performance, highlighting areas for improvement and ensuring evaluations are consistently high-quality.