Skip to content

Latest commit

 

History

History
33 lines (21 loc) · 1.28 KB

README.md

File metadata and controls

33 lines (21 loc) · 1.28 KB
Raw

TreeCaps: Tree-based Capsule Networks for Source Code Processing

This repository contains the code for TreeCaps introduced in the following paper TreeCaps: Tree-based Capsule Networks for Source Code Processing. (AAAI 2021)

Usage

  1. Install the required dependencies pip install -r requirements.txt.

  2. Download and extract the dataset python3 download_data.py. After this step, you can see OJ_data folder, noted that this data has been parsed into SRCML-based AST (https://www.srcml.org/) as protobuf format.

  3. To train the model:

  • Run python3 mains.py --training

Datasets

We used 2 datasets in 2 programming languages to ensure cross-language robustness:

  • [SA Dataset:]: 10 classes of sorting algorithms, with 64 training programs on average per class, written in Java.
  • [OJ Dataset:]: 104 classes of C programs, with 375 training programs on average per class.

Citation

If you find this tutorial useful for your research, please consider citing our paper:

@article{bui2020treecaps,
  title={TreeCaps: Tree-Based Capsule Networks for Source Code Processing},
  author={Bui, Nghi DQ and Yu, Yijun and Jiang, Lingxiao},
  journal={arXiv preprint arXiv:2009.09777},
  year={2020}
}