aabbas90 / adda-pytorch Goto Github PK

Hi Thanks for your work.

When I run Phase2MNISTtoUSPS.py, I find the result to be wrong:

Test Accuracy of the model on the source test images: 99 %
Test Accuracy of the model on the target test images: 1445 %
Train Itr: 1 D Loss: 0.704938 Target Loss: 0.678370

Accuracy of target on target validation images: 1445 %
Accuracy of D on validation images: 51 %
Train Itr: 101 D Loss: 0.680300 Target Loss: 0.699642

Train Itr: 201 D Loss: 0.687932 Target Loss: 0.704488

Train Itr: 301 D Loss: 0.672281 Target Loss: 0.710689

Accuracy of target on target validation images: 1447 %
Accuracy of D on validation images: 59 %
Train Itr: 401 D Loss: 0.682694 Target Loss: 0.712692

Train Itr: 501 D Loss: 0.666517 Target Loss: 0.720006

Max target accuracy achieved is 1447 %
Test Accuracy of the model on the source test images: 75 %
Test Accuracy of the model on the target test images: 1444 %
Test Accuracy of the model on the target train images: 1445 %

Is it a bug in your source code?

Hi, thanks for your project.
I run the Phase1SVHN.py and Phase2SVHNtoMNIST.py with your parameters, and only get a 46% accuracy of target on target test image.

Thanks for you code.
I have a Error. Can you help?

Traceback (most recent call last):
  File "Phase1USPS.py", line 130, in <module>
    print('Test Accuracy of the model on the test images: %d %%' % (100 * correct / total))
ZeroDivisionError: division by zero