This repository contains code of research project "Auto-Generated Code Detection".
Repository structure:
- code2vec -- third party implementation of code2vec model based on this repository
- dataset -- Java source code for automatic collection of dataset and features calculation
- git_python_wrapper -- python wrapper for GitHib search API
- experiments -- source code of experiments
Folder experiments/datasets
contains datasets for both simple features evaluation and code2vec features evaluation.
Folder experiments/training/*
contains authors evaluation results.
To rerun all experiments with presented data run python experiments/training/train.py
(works several hours for all models)
Final presentation link: https://docs.google.com/presentation/d/1MbK6Xdf2J9RknKyqMrSGeKLb5s1KEI32RY3IMbKV864/edit?usp=sharing