Saved in:
Bibliographic Details
Main Authors: Chen, Dong, Lin, Shaoxin, Zeng, Muhan, Zan, Daoguang, Wang, Jian-Gang, Cheshkov, Anton, Sun, Jun, Yu, Hao, Dong, Guoliang, Aliev, Artem, Wang, Jie, Cheng, Xiao, Liang, Guangtai, Ma, Yuchi, Bian, Pan, Xie, Tao, Wang, Qianxiang
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.01304
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913384806481920
author Chen, Dong
Lin, Shaoxin
Zeng, Muhan
Zan, Daoguang
Wang, Jian-Gang
Cheshkov, Anton
Sun, Jun
Yu, Hao
Dong, Guoliang
Aliev, Artem
Wang, Jie
Cheng, Xiao
Liang, Guangtai
Ma, Yuchi
Bian, Pan
Xie, Tao
Wang, Qianxiang
author_facet Chen, Dong
Lin, Shaoxin
Zeng, Muhan
Zan, Daoguang
Wang, Jian-Gang
Cheshkov, Anton
Sun, Jun
Yu, Hao
Dong, Guoliang
Aliev, Artem
Wang, Jie
Cheng, Xiao
Liang, Guangtai
Ma, Yuchi
Bian, Pan
Xie, Tao
Wang, Qianxiang
contents GitHub issue resolving recently has attracted significant attention from academia and industry. SWE-bench is proposed to measure the performance in resolving issues. In this paper, we propose CodeR, which adopts a multi-agent framework and pre-defined task graphs to Repair & Resolve reported bugs and add new features within code Repository. On SWE-bench lite, CodeR is able to solve 28.33% of issues, when submitting only once for each issue. We examine the performance impact of each design of CodeR and offer insights to advance this research direction.
format Preprint
id arxiv_https___arxiv_org_abs_2406_01304
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CodeR: Issue Resolving with Multi-Agent and Task Graphs
Chen, Dong
Lin, Shaoxin
Zeng, Muhan
Zan, Daoguang
Wang, Jian-Gang
Cheshkov, Anton
Sun, Jun
Yu, Hao
Dong, Guoliang
Aliev, Artem
Wang, Jie
Cheng, Xiao
Liang, Guangtai
Ma, Yuchi
Bian, Pan
Xie, Tao
Wang, Qianxiang
Computation and Language
Artificial Intelligence
Software Engineering
GitHub issue resolving recently has attracted significant attention from academia and industry. SWE-bench is proposed to measure the performance in resolving issues. In this paper, we propose CodeR, which adopts a multi-agent framework and pre-defined task graphs to Repair & Resolve reported bugs and add new features within code Repository. On SWE-bench lite, CodeR is able to solve 28.33% of issues, when submitting only once for each issue. We examine the performance impact of each design of CodeR and offer insights to advance this research direction.
title CodeR: Issue Resolving with Multi-Agent and Task Graphs
topic Computation and Language
Artificial Intelligence
Software Engineering
url https://arxiv.org/abs/2406.01304