Implementation of the paper Sketch and Customize: A Counterfacutal Story Generator .

1. Preparation.

pip install -r requierements.txt

2. Get the labels of the causal and background words, and the various types of skeletons.

python get_augment_skeletons.py

3. Training of the sketch model.

bash sketch_train.sh

4. Inference of the sketch model, results will be saved in sketch_pred_results dir.

bash sketch_inference.sh

5. Training of the customize model.

bash customize_train.sh

6. Inference of the customzie models with various types of skeletons.

bash customize_inference.sh

7. Evaluation of the models.

python evaluation.py