mnist.py Bug

[bhqt]

When I am run mnist.py then
TypeError("int() argument must be a string, a bytes-like object or a number, not 'NoneType'",)

[gsantopaolo]

Got the same bug every time at step 8500, any idea what's going on there?

I’m running VS 2017 15.8.2 + AI Tools on Windows 1803.
I was also able to install:

• tensorflow-gpu
• CUDA 9.0.176
• cuDNN 7.0.5 for CUDA 9.0

Every time I try to train the model out of mnist.py I always got an error at step 8500
Stack trace:

File "c:\dev\mnist\mnist.py", line 409, in
tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)

I’ve already tried:

• deleting the input dir so that the app downloads again the train/test sets
• downloaded train/test sets with a browser
• deleting the output dire

I’ve also reported the issue on the git repo here.
Any help is well accepted

Thank you

[TobeyQin]

@gsantopaolo Hi, thank you for your report. Could you tell me Which Python interpreter and version you installed? CPython or Anaconda? 3.5 or 3.6? And which version of tensorflow you installed?

I can run the script successfully with below configuration:
Python env: python3.5
tf version: tensorflow-gpu==1.5.0
CUDA: CUDA 9.0.176
cuDNN: 7.0.5

As a workaround, maybe you can uninstall and reinstall tensorflow-gpu, and delete the input and output dir, then have another try.

Thanks.

[gsantopaolo]

running cpython 3.6

and tesorflow-gpu 1.5.0

Have already fighted with tensorflow-gpu unistalled and reinstalled several times

It seems the only difference we have is python you are running 3.5 and i'm runnig 3.6

@TobeyQin Can you test if it works for you with python 3.6?

[gsantopaolo]

@TobeyQin I've just isntalled a new CPython 3.5.4 environment on Visual Studio 2017.
An I can confirm that i'm experiencing the same error

Message=SystemExit()
Source=c:\dev\samples-for-ai-master\examples\tensorflow\mnist\mnist.py
StackTrace:
File "c:\dev\samples-for-ai-master\examples\tensorflow\mnist\mnist.py", line 409, in
tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)

[gsantopaolo]

So I understood where there error is happening, unfortunately I didn't understood why so I wasn' t able to find a solution.
Basically the app is running perfectly until the end, even the pb model is saved on the export dir.
But when the code hits line 374
if FLAGS.self_test:
it fires an exception
No matter if self_test is true or false (you can configure it from the project properties-->script arguments)

[TobeyQin]

I can run with python3.6 provided by visual studio python tools. I cannot reproduce your bug.
Please email to tools4AI_feedback@microsoft.com, then I will ping you later.

[gsantopaolo]

@TobeyQin i tried to mail you but it seems noone ouside microsoft is allowe to send mail to that address...

[linmajia]

@gsantopaolo , we corrected the setting and please try again.

Thanks.