0

I have saved the training in a checkpoint then i restored it in order to add one more label after that when i trying to run the code it always repeat the training. is there any syntax or flags to know if it's trained or not ?

peter
  • 1
  • 5

1 Answers1

0

We'd need to see how you're restoring it. It sounds like you're either not restoring it properly or you are reseting the weights once they have been restored.

There's no way to know if it's "trained" the only thing you can check for is if it got restored properly. If it doesn't get restored it will usually throw an error.

Steven
  • 5,134
  • 2
  • 27
  • 38
  • that's the link of my code http://stackoverflow.com/questions/34913762/how-to-expand-a-tensorflow-variable , i already posted there and no body answered me :( hopefully if you have answer for that question also – peter Sep 30 '16 at 14:44
  • is that the right link? How to expand a Tensorflow Variable? That's the title? There's no code there either and that question was marked as solved – Steven Sep 30 '16 at 15:38
  • oh sry man that's the link http://stackoverflow.com/questions/39746542/expend-the-output-of-cnn it's like the same name that why i send it to you without intention – peter Sep 30 '16 at 16:04
  • the problem is here for step in xrange(num_steps): sess.run(init_op) Why do you do this? get rid of it. – Steven Sep 30 '16 at 16:07
  • I posted an answer on the other one. – Steven Sep 30 '16 at 16:09