LSTM Cannot Convert Tensor To Numpy Array (Not Implemented Error)

Question

I am trying to build an LSTM to predict the next word in a sentence. But I keep getting an error:

NotImplementedError: Cannot convert a symbolic Tensor (lstm/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

The only code I've put in so far is this (excluding the imports of libraries)

model=Sequential()
model.add(LSTM(100,input_shape=(5,3590),return_sequences=True))
model.add(BatchNormalization())
model.add(LSTM(50,return_sequences=True))
model.add(BatchNormalization())
model.add(LSTM(25))
model.add(BatchNormalization())
model.add(Flatten())
model.add(Dense(26,activation="softmax"))

I haven't compiled or fitted the model left. I know there is a way to solve this simply by switching either your TensorFlow/Numpy version or your Python version but I was wondering if there was a way to solve this without having to change versions.

Any help would be appreciated.

Answer 1

Try downgrading to numpy 1.19.5

pip install numpy==1.19.5