Associate Professor
Shanghai Jiao Tong University
Professor
Shanghai Jiao Tong University
Abstract
The emergence of language model changes the way of how we program. In the long run, programmers will write, edit, test, debug, and repair code with code agents. This workshop covers the cutting-edge topic on how AI and software engineering research is developed to generate, edit, test, and repair the code. By walking through all techniques to automate programmers' work, we aim to forsee how auto-programming techniques can advance our life in the upcoming 5-10 years.
Organizers
Speakers
(Click to view each speaker's talk information. Click on the name to visit the homepage.)
CoEdPilot: Recommending Code Edits with Learned Prior Edit Relevance, Project-wise Awareness, and Interactive Nature
Speaker
Chenyan Liu
Abstract
Incremental code edits are more frequent than generating new code in software projects. To automate this process, existing language model-based solutions focus primarily on generating edit solutions based on given location and relevant prior edits. However, editing tasks can be more complicated: It is non-trivial to infer the subsequent edit location, as the scope of edit ripple effect can be the whole project. Moreover, editing sessions may contain multiple (ir)relevant edits. In this talk, I will share our work CoEdPilot, a LLM-driven framework for code evolution assistance. CoEdPilot orchestrates a set of neural Transformers, for discriminating relevant edits, monitoring the ripple effects of edits, exploring their interactive natures and generating edit solution. We also implement CoEdPilot as a VS Code extension for user-friendly interaction.
Bio
Chenyan Liu is currently a PhD student at School of Computing, National University of Singapore. He is supervised by Dr. Yun LIN, Dr. Jin Song DONG and Dr. ZhiYong HUANG. He received his Bachelor’s degree at Huazhong University of Science and Technology in 2021 and Master’s degree at National University of Singapore in 2023. His research focuses on the design and evaluation of code evolution systems.
Location
Singapore Management University,
Administration Building, Room 4-1
Agenda
December 18th, 2024
| Time | Speaker | Duration | Topic |
|---|---|---|---|
| 3:00PM - 3:40PM | 20min + 20min | ||
| 3:40PM - 4:20PM | |||
| 4:20PM - 5:00PM | |||
| 5:00PM - 5:10PM | 10min | ||
| 5:10PM - 5:45PM | 15min + 20min | ||
| 5:45PM - 6:20PM |