Skip to content
Implementation of "Automatic Source Code Summarization with Extended Tree-LSTM"
Python Jupyter Notebook
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
notebooks Add check_result notebook Dec 2, 2018
parser
README.md Update README.md Jul 2, 2019
dataset.py Change savedata name & Change nl first charactor to lower Dec 2, 2018
layers.py Fasten BiLSTM with CuDNN Nov 19, 2018
models.py Remove unused slice Nov 28, 2018
requirements.txt Fix bug Sep 21, 2018
retrain.py Not shuffle data Nov 22, 2018
train.py Add explicit data sorting (the move is same) Dec 3, 2018
utils.py Fix bug (dropout was always .5) Nov 22, 2018

README.md

Attention-based Tree-to-Sequence Code Summarization Model

The TensorFlow Eager Execution implementation of Source Code Summarization with Extended Tree-LSTM (Shido+, 2019)

including:

  • Multi-way Tree-LSTM model (Ours)
  • Child-sum Tree-LSTM model
  • N-ary Tree-LSTM model
  • DeepCom (Hu et al.)
  • CODE-NN (Iyer et al.)

Dataset

  1. Download raw dataset from [https://github.com/xing-hu/DeepCom]
  2. Parse them with parser.jar

Usage

  1. Prepare tree-structured data with dataset.py
    • Run $ python dataset.py [dir]
  2. Train and evaluate model with train.py
    • See $ python train.py -h
You can’t perform that action at this time.