Reinforcement learning to improve large language model-based automated code compliance systems

2026-06-21 · Jack Wei Lun Shi, Minghao Dang, Wawan Solihin, Leong Hien Poh, Justin K. W. Yeoh arxiv

Large language model (LLM)-based approaches for automated code compliance (ACC) of building regulations are prone to generating incorrect and hallucinated computer-processable rules. This paper introduces P4IR, a two-stage framework that uses supervised fine-tuning (SFT) to instill domain knowledge in an LLM, followed by Group Relative Policy Optimization (GRPO) to improve the accuracy of the generated intermediate representations in the form of high-level code skeletons. The framework achieved reductions of up to 23.8% and 38.6% in tree edit distance and token-level Levenshtein distance respectively, relative to the SFT baselines. Comparative analysis demonstrates that this approach in a zero-shot setting outperforms leading LLMs in both code structure and semantics, specifically Claude Opus and Sonnet 4.5, GPT-5.2, Qwen-3-Max, and GLM-4.7, evaluated via few-shot prompting. Additionally, the GRPO stage produced a small yet statistically significant reduction in false positives. By combining SFT with GRPO to optimize directly for domain-specific objectives, this approach offers a path toward more accurate and reliable LLM-based ACC systems.

📄 PDF Abstract BibTeX arXiv:2606.22402

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

ACECODER: Acing Coder RL via Automated Test-Case Synthesis

2025-02-03 · Huaye Zeng, Dongfu Jiang, Haozhe Wang, Ping Nie 외

Most progress in recent coder models has been driven by supervised fine-tuning (SFT), while the potential of reinforcement learning (RL) remains largely unexplored, primarily due to the lack of reliable reward data/model…

HumanEval mbpp reinforcement-learning Reinforcement Learning+1

LangProp: A code optimization framework using Large Language Models applied to driving

2024-01-18 · Shu Ishida, Gianluca Corrado, George Fedoseev, Hudson Yeo 외

We propose LangProp, a framework for iteratively optimizing code generated by large language models (LLMs), in both supervised and reinforcement learning settings. While LLMs can generate sensible coding solutions zero-s…

Autonomous Driving Code Generation Imitation Learning reinforcement-learning+1

An Encoder-Decoder Based Audio Captioning System With Transfer and Reinforcement Learning

2021-08-05 · Xinhao Mei, Qiushi Huang, Xubo Liu, Gengyun Chen 외

Automated audio captioning aims to use natural language to describe the content of audio data. This paper presents an audio captioning system with an encoder-decoder architecture, where the decoder predicts words based o…

Audio captioning Decoder reinforcement-learning Reinforcement Learning+2

FALCON: Feedback-driven Adaptive Long/short-term memory reinforced Coding Optimization system

2024-10-28 · Zeyuan Li, Yangfan He, Lewei He, Jianhui Wang 외

Recently, large language models (LLMs) have achieved significant progress in automated code generation. Despite their strong instruction-following capabilities, these models frequently struggled to align with user intent…

Code Generation HumanEval Instruction Following mbpp+3

Reinforcement Learning from Automatic Feedback for High-Quality Unit Test Generation

2024-12-18 · Benjamin Steenhoek, Michele Tufano, Neel Sundaresan, Alexey Svyatkovskiy

Software testing is a crucial but time-consuming aspect of software development, and recently, Large Language Models (LLMs) have gained popularity for automated test case generation. However, because LLMs are trained on …

software testing