cchio / deep-pwning Goto Github PK

Here is how I am running it.

(deep-pwning) ttys000:1:[email protected]:[~/Documents/deep-pwning/dpwn]: frameworkpython cifar10_driver.py --test
/Users/joshuaherman/.virtualenvs/deep-pwning/lib/python2.7/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.
  warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')
Filling queue with 20000 CIFAR images before starting to train. This will take a few minutes.
2016-08-14 12:54:12.638315: step 0, loss = 4.67 (23.3 examples/sec; 5.504 sec/batch)
2016-08-14 12:54:19.990925: step 10, loss = 4.65 (196.6 examples/sec; 0.651 sec/batch)

I have ran parts of mnist_driver.py replacing tensorflow with version 0.9 and there are no runtime errors. Once I get to my home computer with a GPU I will compare do some runs of it against 0.8 on my home computer.

If I run CIFAR10 doing "python cifar10driver.py" it doesn't output anything to the folder output .

I attempted to install this on mac os x and looking at the docs for tensorflow you have to choose a tensorflow binary. See https://www.tensorflow.org/versions/r0.8/get_started/os_setup.html . I have fixed it in #2

Hello,
This is a cool project! I like your talk on defcon. But I meet some problem of the code.
I ran mnist_driver.py and generate the mnist pickle file.
I try to use mnist_read_pickle.py, which is in the util folder, to generate the ".png" image.
But it seem like the file mnist_read_pickle.py is not working.

Here is the error I got.

usage: python mnist_read_pickle.py <pickle_file_path> [num_samples_to_persist]
Traceback (most recent call last):
  File "mnist_read_pickle.py", line 77, in <module>
    main(sys.argv)
  File "mnist_read_pickle.py", line 48, in main
    pkl_path = argv[1]
IndexError: list index out of range

Do you have any idea? Thanks!

Hi @cchio,

Big fan of your work and looking forward to your new book! I forked this repository and made it Py3 and Tensorflow 1.4.0 compatible (very easy, thanks for the clean code!). Just wanted to let you know in case you ever wanted a PR to do so.

Would also love to hear if you had any other ideas I could contribute to the project. I will be presenting it at Congress this year. 😄

Cheers,
kjam