Deep code comment generation

• paper: https://xin-xia.github.io/publication/icpc182.pdf
• official code: null

Environment

conda create -n deepcom_env python=3.6 
conda activate deepcom_env
conda install pytorch-gpu=1.3.1
pip install git+https://github.com/casics/spiral.git
pip install nltk==3.2.5  ipdb==0.13.3 javalang==0.12.0 networkx==2.3 prettytable

Data processing

Step1 get data

cd data/tlcodesum
bash get_data.sh

Step2 code and summary processing

cd data/tlcodesum
bash preprocess.sh

Run

cd code/
bash run_tlcodesum.sh $gpus $seed

For example:

bash run_tlcodesum.sh 0 0

Evaluate

refs_filename=../saved_model/test.gold
preds_filename=../saved_model/test.pred
python evaluate.py --refs_filename $refs_filename --preds_filename $preds_filename 

Result

Best epoch	BLEU-DM	GPU
98	40.1778	v100