Xzh0u/CQIL

CQIL

Environment Preparation

conda create -n cqil python=3.7
conda activate cqil
pip install -r requirements.txt

Dataset

There is a data demo in data/example. config.py defines the dataset path. To preprocess the data, you should run

python pipeline.py
# or contains detailed data path
python pipeline.py --data_dir './data/example/' --train_file "train.data_origin.json" --valid_file "valid.data_origin.json" --eval_file "eval.json" > log/pipeline.log

Input: train.data.json and valid.data.json

Train & Evaluate

To train our model:

python main.py --mode train 
# or contains detailed data path
python main.py --mode train --data_dir './data/example/' --train_file "train.data_origin.json" --valid_file "valid.data_origin.json" --eval_file "eval.json"

To evaluate our model:

python main.py --mode eval