AI and LLMs · Middle · Question 66 of 75
What is the difference between code-graded and model-graded evals?
- 1Code checks exact rules; a model grades open qualities
- 2Model-graded evals don't need any test cases at all
- 3Code-graded evals are always written by AI
Show the answer
Correct answer: Code checks exact rules; a model grades open qualities
Explanation
Use code where possible (JSON valid, contains ID); use a judge for tone, completeness or helpfulness.