Incorporating bidirectional interactive information and ...
2021-1-1u2002·u2002We can see that when hidden size is set to 1000, the model has highest performance in two datasets. What's more, the F1 score gets better and better as the hidden size gets higher and higher except the GRU-based model in NYT dataset. Also we can see that LSTM-based model has higher F1 score than its GRU counterpart in this experiment.
Get Price