Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
July 2, 2026Proceedings of the ACM on software engineering.Open Access

Reward-Free Code Alignment from Pretrained or Fine-Tuned LLM: Unpacking the Trade-offs for Code Generation

View Full Paper
Ask AI
Bookmark
Share

Authors

SSSanjeepan SivapiranGUGias Uddin

Discussion

Loading...

Member takes

Overview

Empirical study shows alignment techniques improve code quality in LLMs, suggesting key trade-offs for developers.

Key Points

  • This research investigates the effectiveness of LLM alignment techniques for code generation tasks, focusing on functional and non-functional requirements.
  • Examined five state-of-the-art LLMs using Direct Preference Optimization and BoNBoN techniques.
  • Evaluated functional requirements with four benchmarks and non-functional requirements with the CODAL benchmark.
  • Conducted analyses on both pretrained and finetuned versions of the models.
  • Pretrained-to-aligned pathways show larger improvements (+75% non-functional, +42% functional).
  • Pretrained variant generally less accurate than finetuned variant, but alignment narrows the performance gap.
  • Non-functional requirements consistently improved more than functional through alignment.

Cite This Study

Sivapiran et al. (2026) studied this question.

synapsesocial.com/papers/6a4600489ed1343031310781https://doi.org/10.1145/3808123
View Full Paper
Ask AI
Bookmark
Share