Reproduction Package for Article 'ReDef: Do Code Language Models Truly Understand Code Changes for Just-in-Time Software Defect Prediction?'

This repository contains the official replication package for the paper "ReDef: Do Code Language Models Truly Understand Code Changes for Just-in-Time Software Defect Prediction?" accepted at FSE 2026. (DOI: https://doi.org/10.1145/3808179)The artifact is designed to support the reproducibility of our experimental results and to facilitate the reuse of the ReDef dataset in future research.The package is organized as follows:- ReDef.pdf: A copy of the accepted manuscript.- ReDef.zip: The complete curated dataset, consisting of 3,164 defective and 10,268 clean function-level modifications across 22 large-scale C/C++ projects.- code_artifacts.zip: All experimental scripts for dataset collection and CLM evaluation.- README.txt: Overview of the artifact scheme and data provenance.- INSTALL.txt: Detailed, step-by-step instructions for environment setup and replication.- REQUIREMENTS.txt: Hardware/software requirements.- STATUS.txt: Document stating the claims for Functional, Reusable, and Available badges.- LICENSE.txt: Open-source license for distribution. Instant Setup via Docker:A pre-configured Docker image is available on Docker Hub for out-of-the-box execution:Image: waroad1/redef:fse2026