I'm trying to concatenate two hidden states of a 2-Bi-LSTM layers
but i want to know first what is the output shape of a concatenation.
Moreover, I would like to know what are the attention mechanisme. To be more explicit how can i do it for two different Bi-LSTM hidden states