This project is a hybird of a graph-based and a transition-based parser. The main principle comes from the paper Simple and Accurate Dependency Parsing Using Bidirectional LSTM Feature Representations.
Our repository is forked from elikip's repo. He wrote the code with DyNet. We use Python 3.6 with PyTorch, and my cooperator xiezhq-hermann rewrites the graph-based parser using Python3.
- Python 3.6
- PyTorch >= 0.2.0
- (https://github.com/clab/dynet/tree/master/python)
This software is released under the terms of the Apache License, Version 2.0.