Reputation: 55
While executing udc_train.py
of the chatbot-retrieval project from GitHub, I got this exception at the end of training steps and at the beginning of evaluations step. The problem seems that the following line of code returns None
to the variable _set_ops
:
_set_ops = loader.load_op_library(
resource_loader.get_path_to_datafile("_set_ops.so"))
I could found no _set_ops.so
or so in tensorflow
library folder.
Any suggestions to resolve this problem?
My runtime: Amazon EC2 GPU Windows Instance.
ANSWER: Submitted this ticked to the github community of tensorflow and a new release has been build as a fix.
(I couldn't submit my answer since due to lack of StackOverflow badges. Anyone who is volunteer to submit this answer is more than welcome. I will approve it as the answer.)
Upvotes: 0
Views: 471
Reputation: 21917
OP found answer: This was a bug in TensowFlow that has now been fixed. All 1.0 and later builds of TensorFlow have this error corrected.
Upvotes: 1