Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2509.25243 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908566775922688 |
|---|---|
| author | Tang, Xunzhu Olatunji, Iyiola Emmanuel Sun, Tiezhu Klein, Jacques Bissyande, Tegawende F. |
| author_facet | Tang, Xunzhu Olatunji, Iyiola Emmanuel Sun, Tiezhu Klein, Jacques Bissyande, Tegawende F. |
| contents | LLMs demonstrate surface-level fluency in code generation but struggle with structured reasoning tasks requiring correctness and semantic alignment. While Chain-of-Thought (CoT) prompting enhances reasoning through intermediate steps, it suffers from verbosity and inefficiency. Chain-of-Draft (CoD) prompting offers more concise reasoning, but the stochastic nature of LLMs produces varying solution quality, making optimal selection challenging. We propose \multicod, a reinforcement learning framework that learns to select the most promising candidate from CoD-generated solutions. Our approach uses strategy-guided prompting to encourage diverse reasoning styles and models solution selection as a contextual bandit problem. The framework optimizes interpretable features including code complexity, reasoning structure, and strategic metadata through a reward function balancing correctness, efficiency, and clarity. Experiments on MBPP, BigCodeBench, SWE-bench Verified, and Defects4J show \multicod~outperforms and in some cases, on par with standard prompting, CoT, and CoD baselines while achieving cost and token efficiency from the user's perspective through a multi-candidate design that charges only for the selected output, reducing user billing by over 50\% and improving LLM response quality, making \multicod~more sustainable and scalable for real-world deployment. Our code is available: https://anonymous.4open.science/r/MultiCoD. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_25243 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Reinforcement Learning-Guided Chain-of-Draft for Token-Efficient Code Generation Tang, Xunzhu Olatunji, Iyiola Emmanuel Sun, Tiezhu Klein, Jacques Bissyande, Tegawende F. Software Engineering Artificial Intelligence LLMs demonstrate surface-level fluency in code generation but struggle with structured reasoning tasks requiring correctness and semantic alignment. While Chain-of-Thought (CoT) prompting enhances reasoning through intermediate steps, it suffers from verbosity and inefficiency. Chain-of-Draft (CoD) prompting offers more concise reasoning, but the stochastic nature of LLMs produces varying solution quality, making optimal selection challenging. We propose \multicod, a reinforcement learning framework that learns to select the most promising candidate from CoD-generated solutions. Our approach uses strategy-guided prompting to encourage diverse reasoning styles and models solution selection as a contextual bandit problem. The framework optimizes interpretable features including code complexity, reasoning structure, and strategic metadata through a reward function balancing correctness, efficiency, and clarity. Experiments on MBPP, BigCodeBench, SWE-bench Verified, and Defects4J show \multicod~outperforms and in some cases, on par with standard prompting, CoT, and CoD baselines while achieving cost and token efficiency from the user's perspective through a multi-candidate design that charges only for the selected output, reducing user billing by over 50\% and improving LLM response quality, making \multicod~more sustainable and scalable for real-world deployment. Our code is available: https://anonymous.4open.science/r/MultiCoD. |
| title | Reinforcement Learning-Guided Chain-of-Draft for Token-Efficient Code Generation |
| topic | Software Engineering Artificial Intelligence |
| url | https://arxiv.org/abs/2509.25243 |