Closed nottombrown closed 5 years ago
This should help out with #44 and #46. It runs in the graph and solves memory issues by using tf.map_fn
tf.map_fn
You can see the corresponding commit in https://github.com/tensorflow/cleverhans/pull/623
cc @yenchenlin and @carlini
It requires a new version of Cleverhans to be released before we can put the correct requirements in the setup.py file. For the meantime, I've just copied those changes in here and will remove them when the new Cleverhans is released
Thanks, this will be very helpful.
This should help out with #44 and #46. It runs in the graph and solves memory issues by using
tf.map_fn
You can see the corresponding commit in https://github.com/tensorflow/cleverhans/pull/623
cc @yenchenlin and @carlini
It requires a new version of Cleverhans to be released before we can put the correct requirements in the setup.py file. For the meantime, I've just copied those changes in here and will remove them when the new Cleverhans is released