TCSE logo 
 Sigsoft logo
Sustainability badge
Sat 3 May 2025 10:10 - 10:20 at 214 - Opening / Keynote 1 / Paper Session 1 Chair(s): Zijian Wang

Large Language Models (LLMs) have become integral to various software engineering tasks, including code generation, bug detection, and repair. To evaluate model performance in these domains, numerous bug benchmarks containing real-world bugs from software projects have been developed. However, a growing concern within the software engineering community is that these benchmarks may not reliably reflect true LLM performance due to the risk of data leakage. Despite this concern, limited research has been conducted to quantify the impact of potential leakage.

In this paper, we systematically evaluate popular LLMs to assess their susceptibility to data leakage from widely used bug benchmarks. To identify potential leakage, we use multiple metrics, including a study of benchmark membership within commonly used training datasets, as well as analyses of negative log-likelihood and n-gram accuracy. Our findings show that certain models, in particular CodeGen, exhibit significant evidence of memorization in widely used benchmarks like Defects4J, while newer models trained on larger datasets like LlaMa 3.1 exhibit limited signs of leakage. These results highlight the need for careful benchmark selection and the adoption of robust metrics to adequately assess models capabilities.

Sat 3 May

Displayed time zone: Eastern Time (US & Canada) change

09:00 - 10:30
Opening / Keynote 1 / Paper Session 1LLM4Code at 214
Chair(s): Zijian Wang AWS AI Labs
09:00
10m
Day opening
Opening
LLM4Code
Lingming Zhang University of Illinois at Urbana-Champaign, Prem Devanbu University of California at Davis, Zijian Wang AWS AI Labs
09:10
60m
Keynote
Keynote 1: Building the Hybrid Human-AI Developer: From Code Completion to Agents (zoom talk)
LLM4Code
10:10
10m
Talk
Are Large Language Models Memorizing Bug Benchmarks?
LLM4Code
Daniel Ramos Carnegie Mellon University, Claudia Mamede Carnegie Mellon University, Kush Jain Carnegie Mellon University, Paulo Canelas Carnegie Mellon University, Catarina Gamboa Carnegie Mellon University, Claire Le Goues Carnegie Mellon University
10:20
10m
Talk
RepairBench: Leaderboard of Frontier Models for Program Repair
LLM4Code
André Silva KTH Royal Institute of Technology, Martin Monperrus KTH Royal Institute of Technology
:
:
:
: