Closed WMF1997 closed 4 years ago
perhaps it is the version of tf?
in fact, i am now using tf 1.12.0+ (i.e. the tf version i am now using is more than 1.12.0) (only the server of the lab is using tf 1.8.0, and it is quite high.)
so... i want to change the code. all the operation def
s mentioned in tf_utils. (the defination of a variable on cpu still works fine, and i will not change that)
and i think it will work.
i will try it later in the day.
yours sincerely @wmf1997
I tried my idea in tf 1.13.0 and it finally works. close the issue.
hello @charlesq34
first, thank you for your code!
I want to run the code. And I met a problem:
problem description
I want to test the code in
tf_util.py
, and i picked out theconv2d
part in that.py
file, and I copied and modified the code intransform_nets.py
.and I met the problem here. I think it is the problem of BatchNorm part.
the full description of the error:
my environment:
OS: Ubuntu 16.04 (NO GPU DEVICE) TensorFlow 1.13.1 (CPU) (Is my tf version too high to run it, or not? And I want to run it on tf 1.8.0, in another machine with cuda. I do not know if it could run.)
(the cpu machine is used just for reading your source code)
my added code:
yours sincerely @wmf1997