Regarding duplicated queries in the Square Attack about adversarial-robustness-toolbox HOT 2 OPEN

Hi @Lodour Thank you for using ART! How does you proposed solution with the same functionality look like?

from adversarial-robustness-toolbox.

My workaround is passing y_pred to self.loss, but this only handles the default loss function self._get_logits_diff

The method would become something like

def _get_logits_diff(self, x: np.ndarray, y: np.ndarray, y_pred: Optional[np.ndarray] = None):
    if y_pred is None:
        y_pred = self.estimator.predict(x, batch_size=self.batch_size)
    ...

I haven't come up with a general solution for custom loss functions.

from adversarial-robustness-toolbox.