CodeR: Issue Resolving with Multi-Agent and Task Graphs
Dong Chen, Shaoxin Lin, Muhan Zeng, Daoguang Zan and, Jian-Gang Wang, Anton Cheshkov, Jun Sun, Hao Yu, Guoliang Dong, and Artem Aliev, Jie Wang, Xiao Cheng, Guangtai Liang, Yuchi Ma, and Pan Bian, Tao Xie, Qianxiang Wang
TL;DR
This paper introduces CodeR, a multi-agent system utilizing task graphs to improve bug fixing and feature addition in code repositories, demonstrating promising initial results on SWE-bench lite.
Contribution
The paper presents a novel multi-agent framework with task graphs for issue resolution, advancing automated bug fixing and feature addition techniques.
Findings
CodeR solves 28.33% of issues on SWE-bench lite with a single submission.
Performance analysis of CodeR's design choices offers insights for future improvements.
The approach integrates multi-agent collaboration with structured task planning.
Abstract
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.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Graph Neural Networks · Topic Modeling · Web Data Mining and Analysis