A Systematic Literature Review of Parameter-Efficient Fine-Tuning for Large Code Models
Saima Afrin*, Md. Zahidul Haque*, Antonio Mastropaolo
ACM TOSEM
A systematic literature review synthesizing findings from 28 peer-reviewed papers on Parameter-Efficient Fine-Tuning (PEFT) across a wide range of software-engineering tasks. We analyze how PEFT techniques adapt large deep-learning models by updating only a small subset of parameters, examine their impact on both performance and efficiency, and derive a comprehensive taxonomy that categorizes PEFT usage by task type — distinguishing generative (e.g., code summarization) from non-generative (e.g., code clone detection) scenarios — to guide future research and practical, sustainable deployment. (* denotes equal contribution and joint lead authorship.)