How To Pickle Weakref In Python Tensorflow-keras?

I have written a voice recognition python , i have used tenserflow keras model. When i am giving the input to my model after running for 200 Epoch the exception raises, i tried to

Solution 1:

I had the same issue with saving a Keras model and all of these mentioned methods such as pickle, joblib or even model.save('filename') didn't work. Finally, I got through this by calling directly the save_model method as bellow:

import tensorflow as tf
tf.keras.models.save_model(model, filename)

It may work this way specially if you have trained a tensorflow-keras model with subclasses. Hope it would be helpful to others!!

Solution 2:

You can use joblib or just use model.save('filename.h5)

And to read use from keras.models import load_model

To load model model = load_model('model_trained.h5')

To save model using joblib

joblib.dump('filename.pkl')

And to load

loaded_model = joblib.load('filename.pkl')

for more help on pickle and joblib take a glance here

Solution 3:

It is computationally expensive but you can try running it on 2000 Epochs

Because it can happen that your model is not trained enough

Solution 4:

you can try searching on google by this phrase

Or just try (If yo have a 32 bit python installed open the one with 32 bit)this and then this

And then restart the IDLE