1. Environment

Python 3.6

PyTorch 0.3.1

2. How to run

First, place your train, dev and test data in certain directory and revise the corresponding paths in run_main.sh .

Then, customize your hyper parameters in config.json.

Finally, bash run_main.sh.

This is a multi-purpose piece code, and can be used for training, testing or decoding your own dataset.