• DACL
• How to use
• Datasets and Models
• Acknowledgement
• Reference

DACL

This repository contains code for paper "Bridging Pre-trained Models and Downstream Tasks for Source Code Understanding". (Accepted to the 44th International Conference on Software Engineering (ICSE 2022))

How to use

We apply our approach to code pre-trained models CodeBERT and GraphCodeBERT for fine-tuning.

Please refer to the subdirectory of tasks for more details!

For general fine-tuning (baseline models), please refer to algorithm classification, code clone detection, and code search.

Datasets and Models

We also provide datasets and pre-trained models fine-tuned by our approach to verify the results.

Acknowledgement

Our implementation is adapted from CodeXGLUE and GraphCodeBERT for the implementation of pre-trained models.

Reference

If you use this code, please consider citing us.

@inproceedings{Wang2021BridgingPM,
  title={Bridging Pre-trained Models and Downstream Tasks for Source Code Understanding},
  author={Deze Wang and Zhouyang Jia and Shanshan Li and Yue Yu and Yun Xiong and Wei Dong and Xiangke Liao},
  year={2021}
}