You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I also think that it would be good if SPSA could be called without any explicit labels (the default behavior for other classes is to use the models labels on the clean data as the label to attack in that case.
The text was updated successfully, but these errors were encountered:
I spent some time this afternoon trying to get SPSA to hook into the
generate_np
harness, but had trouble getting it working.Currently the SPSA test works correctly using the
sess.run
syntax, but it would be slightly nicer to have it be callable withgenerate_np
https://github.com/tensorflow/cleverhans/pull/498/files#diff-ee72988b849d43f34dfd258c5f95fc8dR270
I also think that it would be good if SPSA could be called without any explicit labels (the default behavior for other classes is to use the models labels on the clean data as the label to attack in that case.
The text was updated successfully, but these errors were encountered: