2
$\begingroup$

Model works and fits. After adding model.save('model.h5'). I am receiving tensorflow.python.framework.errors_impl.FailedPreconditionError: Attempting to use uninitialized value dense_11/bias[[Node: _retval_dense_11/bias_0_0 = _RetvalT=DT_FLOAT, index=0, _device="/job:localhost/replica:0/task:0/device:CPU:0"]]

I tried model.save_weights('model.h5') saves the model but after trying to load the model I get:

KeyError: "Unable to open object (object 'dense_12' doesn't exist)"

After saving model to json and then trying to load it i get json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0).

What should I do? Thanks for help!

Improve this question
$\endgroup$
2
  • 1
    $\begingroup$ Are you trying to load this saved weights into a newly created network with same architecture and then it is throwing this error? $\endgroup$
    – Kiritee Gak
    Dec 29, 2018 at 7:28
  • $\begingroup$ You need to redefine the architecture exactly in the similar fashion $\endgroup$
    – Aditya
    Dec 29, 2018 at 7:32

1 Answer 1

Reset to default
3
$\begingroup$

The error was caused because after training the model I opened another tensorflow session to save predictions to dataset and put model.save('model.h5') after. So moving this line up, right after the model finished training solved the problem.

Improve this answer
$\endgroup$
1
  • $\begingroup$ you should accept your own answer if that was what solved the problem described in the question. Next time you should also add a snippet of the code you are using to describe what happened in the code so that other users could figure out what was the problem as well. Also, consider adding the lines of code that were the problem and the ones that were the solution. $\endgroup$
    – wacax
    Jan 1, 2019 at 19:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.