/code2vec

A PyTorch implementation of `code2vec: Learning Distributed Representations of Code` (Alon et al., 2018)

Primary LanguagePythonMIT LicenseMIT

PyTorch code2vec

This repo contains an implementation of code2vec: Learning Distributed Representations of Code.

Requirements

  • Python 3+
  • PyTorch 1.0
  • A CUDA compatible GPU

Quickstart

  1. ./download_preprocessed.sh to get the pre-processed datasets from the code2vec and the code2seq papers.
  • Note: by default the 3 datasets from code2seq are commented out.
  1. python run.py

To-Do

  • Graph of results
  • Inference code
  • Embedding exploration