TypeError when run main.py about dcgan

[zsdonghao]

it run well on my side:

• TensorLayer (both 1.4.5 and 1.8.1)
• TensorFlow 1.4.
• Python 3.5
• Ubuntu 16

and I will update this repo with TensorLayer 1.8.1 soon.

[arisliang]

Would you be able to try on TensorFlow 1.6, python 3.6? Wonder if it's tensorflow version, mine is 1.6 and python 3.6.

[TIOPG]

the code in main.py,line22
flags.DEFINE_integer("train_size", np.inf, "The size of train images [np.inf]")
change to :
flags.DEFINE_float("train_size", np.inf, "The size of train images [np.inf]")
then you can run it correct,but there is another tings I don't know you can't got the result you need.
the tutorial in README is sample but unuse...
I suggest you use another sample DCGAN，like:
https://github.com/carpedm20/DCGAN-tensorflow

[arisliang]

I changed the line 22, and it fixed the error, $ python main.py is run successfully. Indeed, directly running main.py looks like training, since it doesn't generate images in the end.

Trying the other repo mentioned above, there're more arguments to the usage of main.py mentioned in README.cmd there, like whether its training, or testing; specifying dataset MNIST or celebA, even your own dataset if have.

[TIOPG]

About this,you can down load the mnist from the page: http://yann.lecun.com/exdb/mnist/ If you know how to translate the Chinese webpage to the English that you can see my blog: http://blog.csdn.net/wei_xiaoyu/article/details/79579829 My English is poor，maybe many words use uncorrect . At 2018-03-17 08:54:40, "Aris L" <[email protected]> wrote: I changed the line 22, and it fixed the error, $ python main.py is run successfully. Indeed, directly running main.py looks like training, since it doesn't generate images in the end. Trying the other repo mentioned above, there're more arguments to the usage of main.py mentioned in README.cmd there, like whether its training, or testing; specifying dataset MNIST or celebA, even your own dataset if have. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.