PyTorch_Biaffine_Dependency_Parsing
PyTorch_Biaffine_Dependency_Parsing copied to clipboard
Hello, can I ask you one question. Why use LSTMCell instead of LSTM in LSTM layer?
LSTM no need to write a loop to do one pass of backprop through time. And num_layers can be set as LSTM parameter. I think it's much easier to use LSTM. Why not use LSTM?