Search code examples
keraslstmrecurrent-neural-networkkeras-layer

Keras - How to get GRU cell state?


For LSTM we can retrieve last output and last cell state as below

outputs, state_h, state_c= LSTM(lstm_dim, return_sequences=True, return_state=True)(inputs)

where state_h is the last sequence of outputs; outputs[-1]=state_h and state_c is the last CELL STATE of LSTM.

But with GRU it does not have such an option but only gives which is equivalent to state_h. How do I get state_c the last cell state from Keras GRU?


Solution

  • There is no internal memory in GRU, which means there is no cell state.

    GRU, in general directly operate on the hidden state to get the output value. For most of the purpose, GRU has the same performance as that of LSTM, while with less computation (less number of parameters).

    Feel free to ask question.