Translation-Invariant-Attacks icon indicating copy to clipboard operation
Translation-Invariant-Attacks copied to clipboard

Published 20 hours ago • verified publisher icon dongyp13

error about squeeze

Open cxmscb opened this issue 5 years ago • 6 comments

''' InvalidArgumentError (see above for traceback): Tried to explicitly squeeze dimension 1 but dimension was not 1: 2 '''

cxmscb avatar May 02 '19 14:05 cxmscb

Sorry, where does the error occur?

dongyp13 avatar May 03 '19 07:05 dongyp13

''' InvalidArgumentError (see above for traceback): Tried to explicitly squeeze dimension 1 but dimension was not 1: 2 '''

This could be a version issue for python and TensorFlow. You can try to change it with the README.

tuji-sjp avatar May 04 '19 12:05 tuji-sjp

I met the same issue. Did you solve it?

lisenxd avatar May 13 '19 08:05 lisenxd

I met the same issue

YuehuaZhu avatar May 24 '19 10:05 YuehuaZhu

I met the same issue. Did you solve it?

solved

lisenxd avatar May 24 '19 10:05 lisenxd

Hi,Yinpeng. a question. why do you set the prediction of inception_resnet_v2 as the target label? and the loss is computed by the (y,pred_logits). so is this means that you want to genarate a adv with target label as inception_resnet_v2 predicted rather than a adv that can fool all the 4 models you gave in the model? if it is convient, hope to get you reply? thx

Ichbinhippo avatar Aug 26 '19 07:08 Ichbinhippo