Term Kind Topic What it is
Adversarial Evaluation practice Model Evaluation & Red-Teaming Deliberately attempting to make a model behave badly, because a probabilistic system with no fixed expected output cannot be verified by conventional testing.