We uploaded the proposed dataset here.
In this folder, we provided resources in two languages, zh (chinese) and en (english).
We provided the code in two allennlp versions, 2.7.0 and 0.9.0, you can choose the version you want.
Firstly, we select the folder corresponding to the version and enter it.
Development environment:
python == 3.6.0
# Requirements
pip install -r requirements.txt
Then, we downloaded the dataset and put it in the directory.
We can edit the specific setting in configuration ''config_for_xxx.json''
allennlp train config_for_xxx.json -s xxx_dir --include-package xxx
the parameter and configuration of trained model will be saved in xxx_dir
We provide evaluate_script.py to evaluate the trained model.
python evaluate_script.py --archive_file xxx1 --include-package xxx2 --language xxx3
"xxx1" is the trained model directory which contain the parameters of model and any other files, "xxx2" is the package which you include in when training. "xxx3" is the language version you choosed.