1 paper
David Troxell, Yulia Alexandr, Sofia Hunt +2
Neural network verifiers aim to provide formal guarantees on model behavior, but existing verification benchmarks are fundamentally limited by their lack of ground-truth labels. As…