TGCRE

Env:

python: 3.9

Dependencies:

Run the following script to install dependencies.

pip3 install -r requirements.txt

File Structure:

The file structure of this repository is shown below.

TGCRE
├── readme.md
├── ckpt
├── data
│   ├── fewrel
│   │   ├── fewrel_dataset.json
│   │   └── relation_description
│   ├── rel2id
│   │   ├── fewrel_rel2id
│   │   └── wikizsl_rel2id
│   └── wikizsl
│       ├── wikizsl_dataset.json
│       └── relation_description
├── model
│   ├── data_process.py
│   ├── evaluation.py
│   ├── inference.py
│   ├── train.py
│   ├── model.py
│   ├── run_fewrel.sh
│   ├── run_inference.sh
│   ├── run_wikizsl.sh
│   └── log
└── requirements.txt

Datasets：

You can download the dataset used in this work via the following google drive link,and then store them in the data/fewrel and data/wikizsl paths respectively.

FewRel (Xu et al., 2018)

WikiZSL (Daniil Sorokin and Iryna Gurevych, 2017)

Train&Inference :

You can easily run training as well as inference with the following scripts

# training stage
# The following two script files differ only in datasets
cd model
bash run_fewrel.sh
bash run_wikizsl.sh

# testing stage
bash run_inference.sh