RNN Models where the output of a cell is the input to the other

TensorFlow
, ,
1

Hi,

I am not very familiar with RNN (LSTM, GRU) models and I wanted to check how I can construct a model where the output of one RNN Cell is the input to the other using Keras.
Do I have to construct my own model using LSTMCell for example?

Thanks

Fadi Badine

1 Like
2

Hi @Fadi_Badine,

In Keras, you can create a sequence-to-sequence model with recurrent layers (e.g., LSTM, GRU) where the output of one cell is used as the input to the next cell within the sequence. You don’t need to create your own LSTM cell; Keras provides high-level APIs to handle this.

Thanks

1 Like
3

These 2 Keras blog on sequence-to-sequence are worth a read:

A ten-minute introduction to sequence-to-sequence learning in Keras

Character-level recurrent sequence-to-sequence model