Title: Bash Comment Generation via Data Augmentation and Semantic-Aware CodeBERT
/data is the experiment corpus.
/result are experiment results.
/train are experiment models.
/GPT is the GPT3.5 model and result.
FGM.py, FSGM.py, NP-GD, and PGD are adversarial training methods.
We have sorted out the code and corpus. You just:
1: Just run "./eval.py" to get the paper results.
2: Just run "./train/run.py" to rerun the experiment.
The README_add includes detailed comparison results and some theories about data augmentation.