Cannot convert a symbolic Tensor (lstm_4/strided_slice:0) to a numpy array.

[Shubham Anand]

x=np.array(np.arange(200))

from tensorflow.keras.models import Sequential

from tensorflow.keras.layers import LSTM, Dense

model = Sequential()

model.add(LSTM(10, activation='relu',input_shape=(look_back,1)))

model.add(Dense(1))

model.compile(optimizer='adam', loss='mse')

num_epochs = 100

model.fit_generator(x, epochs=num_epochs, verbose=1)

NotImplementedError: Cannot convert a symbolic Tensor (lstm_4/strided_slice:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported

Hi, I have posted my code and error , I am getting this error on model.add(LSTM(....))  line. I tried different version of tensorflow and numpy for sorting this error, but Ithe error remains in all the versions.

If we do  not use model.add (LSTM(...)),the  code works properly.

Please help in cleaning this error.

Thanks ,

Shubham

[Peter Wang]

Hi Shubham,

This is a mailing list for conda-related issues, so you'll need to post this to a Tensorflow or Keras mailing list.

-Peter

--
You received this message because you are subscribed to the Google Groups "conda - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to conda+un...@anaconda.com.
To view this discussion on the web visit https://groups.google.com/a/anaconda.com/d/msgid/conda/35077e81-3d8b-4f1f-8649-6db512a0be3cn%40anaconda.com.