LLMs as Continuous Learners: Improving the Reproduction of Defective Code in Software Issues

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Lin, Yalan, Ma, Yingwei, Cao, Rongyu, Li, Binhua, Huang, Fei, Gu, Xiaodong, Li, Yongbin
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915029193850880
author Lin, Yalan
Ma, Yingwei
Cao, Rongyu
Li, Binhua
Huang, Fei
Gu, Xiaodong
Li, Yongbin
author_facet Lin, Yalan
Ma, Yingwei
Cao, Rongyu
Li, Binhua
Huang, Fei
Gu, Xiaodong
Li, Yongbin
contents Reproducing buggy code is the first and crucially important step in issue resolving, as it aids in identifying the underlying problems and validating that generated patches resolve the problem. While numerous approaches have been proposed for this task, they primarily address common, widespread errors and struggle to adapt to unique, evolving errors specific to individual code repositories. To fill this gap, we propose EvoCoder, a multi-agent continuous learning framework for issue code reproduction. EvoCoder adopts a reflection mechanism that allows the LLM to continuously learn from previously resolved problems and dynamically refine its strategies to new emerging challenges. To prevent experience bloating, EvoCoder introduces a novel hierarchical experience pool that enables the model to adaptively update common and repo-specific experiences. Our experimental results show a 20\% improvement in issue reproduction rates over existing SOTA methods. Furthermore, integrating our reproduction mechanism significantly boosts the overall accuracy of the existing issue-resolving pipeline.
format Preprint
id arxiv_https___arxiv_org_abs_2411_13941
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LLMs as Continuous Learners: Improving the Reproduction of Defective Code in Software Issues
Lin, Yalan
Ma, Yingwei
Cao, Rongyu
Li, Binhua
Huang, Fei
Gu, Xiaodong
Li, Yongbin
Software Engineering
Artificial Intelligence
Reproducing buggy code is the first and crucially important step in issue resolving, as it aids in identifying the underlying problems and validating that generated patches resolve the problem. While numerous approaches have been proposed for this task, they primarily address common, widespread errors and struggle to adapt to unique, evolving errors specific to individual code repositories. To fill this gap, we propose EvoCoder, a multi-agent continuous learning framework for issue code reproduction. EvoCoder adopts a reflection mechanism that allows the LLM to continuously learn from previously resolved problems and dynamically refine its strategies to new emerging challenges. To prevent experience bloating, EvoCoder introduces a novel hierarchical experience pool that enables the model to adaptively update common and repo-specific experiences. Our experimental results show a 20\% improvement in issue reproduction rates over existing SOTA methods. Furthermore, integrating our reproduction mechanism significantly boosts the overall accuracy of the existing issue-resolving pipeline.
title LLMs as Continuous Learners: Improving the Reproduction of Defective Code in Software Issues
topic Software Engineering
Artificial Intelligence
url https://arxiv.org/abs/2411.13941