As an example, for an image taken on the ground, two text options24 are: {Several vehicles providing ground transportation are shown in25 the photo: streetcar, tour bus, classic car, and family cars.}

Experiments show that the proposed ReBalanced Adversarial Training (ReBA T) can attain good robustness and does not suffer from robust overfitting even after very long training.

For general structural causal models, we characterize the effect of interventions onso-called stablesets,anotionintroduced by[30].

Ouranalysis include examination of the cross-attention maps of the learned prompts at individual time steps of the denoising process.