What should I do to use Bidirectional LSTM?
Opened this issue · 0 comments
FriedaSmith commented
Hello. I found that Bidirectional LSTM
was not used. Although there is self.bidirectional = kwargs.get("bidirectional", True)
in config.py, it is not used. I added bidirectional: self.lstm = nn.LSTM(self.embedding_length, self.hidden_size, bidirectional=True)
in models.py line: 38, but the following error occurred:
RuntimeError: Expected 3-dimensional tensor, but got 4-dimensional tensor for argument #2 'batch2' (while checking arguments for bmm)
What should I do to use Bidirectional LSTM
?