Experts write the questions. Models answer. Every score checks out.
Test sets written by specialists in their field. Run models on the same set and see every answer behind the score.
Test sets written by specialists in their field. Run models on the same set and see every answer behind the score.