On Pytorch, if you want to build a model like this, ⇓
the code will be:
import torch.nn as nn
BLSTM = nn.LSTM(3, 3, bidirectional=True) # input and hidden sizes are example. it doesn't have to be 3.
But I want to build a model like this. ⇓
This model has two-layerd forward & backward LSTM.
My question is how to build this model in Pytorch.
The code I can think of is this:
import torch.nn
BLSTM = nn.LSTM(3, 3, 2, bidirectional=True) # input and hidden sizes are example. it doesn't have to be 3.
I thought it’s wrong because this ⇓ would be the model if I used this code.
I feel that this model is a little different from the model I want to build.