Don't Transform the Code, Code the Transforms: Towards Precise Code Rewriting using LLMs
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866913541957615616 |
|---|---|
| author | Cummins, Chris Seeker, Volker Armengol-Estapé, Jordi Markosyan, Aram H. Synnaeve, Gabriel Leather, Hugh |
| author_facet | Cummins, Chris Seeker, Volker Armengol-Estapé, Jordi Markosyan, Aram H. Synnaeve, Gabriel Leather, Hugh |
| contents | Tools for rewriting, refactoring and optimizing code should be fast and correct. Large language models (LLMs), by their nature, possess neither of these qualities. Yet, there remains tremendous opportunity in using LLMs to improve code.
We explore the use of LLMs not to transform code, but to code transforms. We propose a chain-of-thought approach to synthesizing code transformations from a small number of input/output code examples that incorporates execution and feedback. Unlike the direct rewrite approach, LLM-generated transformations are easy to inspect, debug, and validate. The logic of the rewrite is explicitly coded and easy to adapt. The compute required to run code transformations is minute compared to that of LLM rewriting.
We test our approach on 16 Python code transformations and find that LLM- generated transforms are perfectly precise for 7 of them and less imprecise than direct LLM rewriting on the others. We hope to encourage further research to improving the precision of LLM code rewriting. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_08806 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Don't Transform the Code, Code the Transforms: Towards Precise Code Rewriting using LLMs Cummins, Chris Seeker, Volker Armengol-Estapé, Jordi Markosyan, Aram H. Synnaeve, Gabriel Leather, Hugh Machine Learning Tools for rewriting, refactoring and optimizing code should be fast and correct. Large language models (LLMs), by their nature, possess neither of these qualities. Yet, there remains tremendous opportunity in using LLMs to improve code. We explore the use of LLMs not to transform code, but to code transforms. We propose a chain-of-thought approach to synthesizing code transformations from a small number of input/output code examples that incorporates execution and feedback. Unlike the direct rewrite approach, LLM-generated transformations are easy to inspect, debug, and validate. The logic of the rewrite is explicitly coded and easy to adapt. The compute required to run code transformations is minute compared to that of LLM rewriting. We test our approach on 16 Python code transformations and find that LLM- generated transforms are perfectly precise for 7 of them and less imprecise than direct LLM rewriting on the others. We hope to encourage further research to improving the precision of LLM code rewriting. |
| title | Don't Transform the Code, Code the Transforms: Towards Precise Code Rewriting using LLMs |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2410.08806 |