Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMs
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917937599741952 |
|---|---|
| author | Yang, Dayu Liu, Tianyang Zhang, Daoan Simoulin, Antoine Liu, Xiaoyi Cao, Yuwei Teng, Zhaopu Qian, Xin Yang, Grey Luo, Jiebo McAuley, Julian |
| author_facet | Yang, Dayu Liu, Tianyang Zhang, Daoan Simoulin, Antoine Liu, Xiaoyi Cao, Yuwei Teng, Zhaopu Qian, Xin Yang, Grey Luo, Jiebo McAuley, Julian |
| contents | In large language models (LLMs), code and reasoning reinforce each other: code offers an abstract, modular, and logic-driven structure that supports reasoning, while reasoning translates high-level goals into smaller, executable steps that drive more advanced code intelligence. In this study, we examine how code serves as a structured medium for enhancing reasoning: it provides verifiable execution paths, enforces logical decomposition, and enables runtime validation. We also explore how improvements in reasoning have transformed code intelligence from basic completion to advanced capabilities, enabling models to address complex software engineering tasks through planning and debugging. Finally, we identify key challenges and propose future research directions to strengthen this synergy, ultimately improving LLM's performance in both areas. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_19411 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMs Yang, Dayu Liu, Tianyang Zhang, Daoan Simoulin, Antoine Liu, Xiaoyi Cao, Yuwei Teng, Zhaopu Qian, Xin Yang, Grey Luo, Jiebo McAuley, Julian Computation and Language Artificial Intelligence Machine Learning Software Engineering In large language models (LLMs), code and reasoning reinforce each other: code offers an abstract, modular, and logic-driven structure that supports reasoning, while reasoning translates high-level goals into smaller, executable steps that drive more advanced code intelligence. In this study, we examine how code serves as a structured medium for enhancing reasoning: it provides verifiable execution paths, enforces logical decomposition, and enables runtime validation. We also explore how improvements in reasoning have transformed code intelligence from basic completion to advanced capabilities, enabling models to address complex software engineering tasks through planning and debugging. Finally, we identify key challenges and propose future research directions to strengthen this synergy, ultimately improving LLM's performance in both areas. |
| title | Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMs |
| topic | Computation and Language Artificial Intelligence Machine Learning Software Engineering |
| url | https://arxiv.org/abs/2502.19411 |