Published September 4, 2022 | Version v3
Software Open

An Empirical Study of Code Smells in Transformer-based Code Generation Techniques

Creators

  • 1. University of Notre Dame
  • 2. Bangladesh University of Engineering and Technology

Description

This repository contains the scripts used for the accepted paper in the research track of the 22nd IEEE International Working Conference on Source Code Analysis and Manipulation (SCAM 2022), titled An Empirical Study of Code Smells in Transformer-based Code Generation Techniques. The paper provides an empirical study on containing (security) code smell in the code generation dataset and the output of the code generation model. It also demonstrates the code smell in the output of GitHub Copilot.

Files

Code_Smell_Artifact.zip

Files (936.0 MB)

Name Size Download all
md5:f074a640df630bf847c8ae33e692dd5a
936.0 MB Preview Download