Skip to main navigation Skip to search Skip to main content

An Empirical Study of Parameter-Efficient Fine-Tuning in Code Change Learning and Beyond

  • Shuo Liu
  • , Jacky Keung
  • , Zhi Jin
  • , Zhen Yang*
  • , Fang Liu
  • , Hao Zhang
  • *Corresponding author for this work
  • City University of Hong Kong
  • Peking University
  • Wuhan University
  • Shandong University

Research output: Contribution to journalArticlepeer-review

Abstract

Compared to Full-Model Fine-Tuning (FMFT), Parameter-Efficient Fine-Tuning (PEFT) has demonstrated superior efficacy and efficiency in several code understanding tasks, owing to PEFT’s ability to alleviate the catastrophic forgetting issue of Pre-trained Language Models (PLMs) by updating only a small number of parameters. However, existing studies primarily involve static code comprehension, aligning with the pre-training paradigm of recent PLMs and facilitating knowledge transfer, but they do not account for dynamic code changes. Thus, it remains unclear whether PEFT outperforms FMFT in task-specific adaptation for code-change-related tasks. To address this question, we examine four prevalent PEFT methods (i.e., AT, LoRA, PT, and PreT) and compare their performance with FMFT across seven popular PLMs. In experiments, two widely studied code-change-related tasks, i.e., Just-In-Time Defect Prediction (JIT-DP) and Commit Message Generation (CMG) are involved, demonstrating that the four PEFT methods can surpass FMFT on JIT-DP but only exhibit comparable performances at best on CMG in common scenarios. While in cross-lingual and low-resource scenarios, they exhibit relative superiority. Afterward, a series of probing tasks from both static and dynamic perspectives are conducted in this paper, offering detailed explanations for the efficacy of PEFT and FMFT. Inspired by the distinctive advantages of PEFT and FMFT in their layer-wise probing results, we propose PastaK , a self-adaPtive efficient layer-specific tuning framework for PLMs in code change learning, which combines FMFT and PEFT during the domain adaptation according to the guidance of probing results. Experiments in the CMG task demonstrate that PastaK surpasses diverse PEFT methods in effectiveness. Even, PastaK outperforms FMFT by 1.48%, 3.21%, and 1.87% at most in terms of BLEU, Meteor, and Rouge-L, while saving 26.26% and 20.65% in terms of training time and computational memory compared with FMFT.

Original languageEnglish
Pages (from-to)3-21
Number of pages19
JournalIEEE Transactions on Software Engineering
Volume52
Issue number1
DOIs
StatePublished - 2026

Keywords

  • Adapter tuning
  • CodeBERT
  • CodeT5
  • CodeT5+
  • GraphCodeBERT
  • PLBART
  • Qwen2.5-Coder
  • UniXcoder
  • commit message generation
  • just-in-time defect prediction
  • low-rank adaptation
  • prefix tuning
  • prompt tuning

Fingerprint

Dive into the research topics of 'An Empirical Study of Parameter-Efficient Fine-Tuning in Code Change Learning and Beyond'. Together they form a unique fingerprint.

Cite this