Proposition: Same clean function and mask, different stress

There are ReLU networks with the same masks and the same clean logits for every input but different stressed accuracies.

Proof. Take $a=1$, isolate a third-layer unit with bias $t>0$, and give its class-$0$ output weight $1/t$. Use competing logits $0,-1,-1$. Clean logits are always $(1,0,-1,-1)$. Under hidden noise, correctness for label $0$ is $1+\sigma G/t>0$, giving

Masks and nonzero-position counts are unchanged as $t$ varies.