Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2505.16901 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918068829028352 |
|---|---|
| author | Tao, Hongyuan Zhang, Ying Tang, Zhenhao Peng, Hongen Zhu, Xukun Liu, Bingchang Yang, Yingguang Zhang, Ziyin Xu, Zhaogui Zhang, Haipeng Zhu, Linchao Wang, Rui Yu, Hang Li, Jianguo Di, Peng |
| author_facet | Tao, Hongyuan Zhang, Ying Tang, Zhenhao Peng, Hongen Zhu, Xukun Liu, Bingchang Yang, Yingguang Zhang, Ziyin Xu, Zhaogui Zhang, Haipeng Zhu, Linchao Wang, Rui Yu, Hang Li, Jianguo Di, Peng |
| contents | Recent advances in Large Language Models (LLMs) have shown promise in function-level code generation, yet repository-level software engineering tasks remain challenging. Current solutions predominantly rely on proprietary LLM agents, which introduce unpredictability and limit accessibility, raising concerns about data privacy and model customization. This paper investigates whether open-source LLMs can effectively address repository-level tasks without requiring agent-based approaches. We demonstrate this is possible by enabling LLMs to comprehend functions and files within codebases through their semantic information and structural dependencies. To this end, we introduce Code Graph Models (CGMs), which integrate repository code graph structures into the LLM's attention mechanism and map node attributes to the LLM's input space using a specialized adapter. When combined with an agentless graph RAG framework, our approach achieves a 43.00% resolution rate on the SWE-bench Lite benchmark using the open-source Qwen2.5-72B model. This performance ranks first among open weight models, second among methods with open-source systems, and eighth overall, surpassing the previous best open-source model-based method by 12.33%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_16901 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Code Graph Model (CGM): A Graph-Integrated Large Language Model for Repository-Level Software Engineering Tasks Tao, Hongyuan Zhang, Ying Tang, Zhenhao Peng, Hongen Zhu, Xukun Liu, Bingchang Yang, Yingguang Zhang, Ziyin Xu, Zhaogui Zhang, Haipeng Zhu, Linchao Wang, Rui Yu, Hang Li, Jianguo Di, Peng Software Engineering Machine Learning Recent advances in Large Language Models (LLMs) have shown promise in function-level code generation, yet repository-level software engineering tasks remain challenging. Current solutions predominantly rely on proprietary LLM agents, which introduce unpredictability and limit accessibility, raising concerns about data privacy and model customization. This paper investigates whether open-source LLMs can effectively address repository-level tasks without requiring agent-based approaches. We demonstrate this is possible by enabling LLMs to comprehend functions and files within codebases through their semantic information and structural dependencies. To this end, we introduce Code Graph Models (CGMs), which integrate repository code graph structures into the LLM's attention mechanism and map node attributes to the LLM's input space using a specialized adapter. When combined with an agentless graph RAG framework, our approach achieves a 43.00% resolution rate on the SWE-bench Lite benchmark using the open-source Qwen2.5-72B model. This performance ranks first among open weight models, second among methods with open-source systems, and eighth overall, surpassing the previous best open-source model-based method by 12.33%. |
| title | Code Graph Model (CGM): A Graph-Integrated Large Language Model for Repository-Level Software Engineering Tasks |
| topic | Software Engineering Machine Learning |
| url | https://arxiv.org/abs/2505.16901 |