Effective auditing metrics are crucial for evaluating the performance of approximate machine unlearning methods.
