Mind the Overlap: Trustworthy Evaluation for Large Code Models

Computer 2026

View at publisher
Authors: Antonio Mastropaolo
DOI: 10.1109/mc.2025.3637575
Cited by: 0
References: 5

Abstract

In this month’s Spotlight on Transactions, we showcase an IEEE Transactions on Software Engineering article by López et al., exploring how hidden data leakage arising from interdataset code duplication artificially elevates benchmark performance and to outline actionable strategies for maintaining rigorous, reliable model assessment.

Fields of Study & Subjects

Software Engineering Research Software Testing and Debugging Techniques Software Engineering Techniques and Practices

Publication details

Journal / source
Computer
Volume
59
Issue
2
Pages
7–9
Publication date
1 February 2026