I have an Enc-Dec LSTM model. I know one way of connecting the two models i.e output of the last state of Enc as Hidden to Dec. But the issue is dimension mismatch.
Encoder output size : [batch_size, seq_len, input_size], ex. [10, 30, 256]
Decoder Hidden size : [num_layer*num_directions, batch_size, hidden_size] ex. [2, 10, 128]
Can dimension issue be resolved? or is there any other way of connecting Enc-Dec together?