Automated Program Repair in the Era of Large Pre-trained Language Models
Open accessAbstract
Automated Program Repair (APR) aims to help developers automatically patch software bugs. However, current state-of-the-art traditional and learning-based APR techniques face the problem of limited patch variety, failing to fix complicated bugs. This is mainly due to the reliance on bug-fixing datasets to craft fix templates (traditional) or directly predict potential patches (learning-based). Large Pre-Trained Language Models (LLMs), trained using billions of text/code tokens, can potentially help avoid this issue. Very recently, researchers have directly leveraged LLMs for APR without relying on any bug-fixing datasets. Meanwhile, such existing work either failed to include state-of-the-art LLMs or was not evaluated on realistic datasets. Thus, the true power of modern LLMs on the important APR problem is yet to be revealed. In this work, we perform the first extensive study on directly applying LLMs for APR. We select 9 recent state-of-the-art LLMs, including both generative and infilling models, ranging from 125M to 20B in size. We designed 3 different repair settings to evaluate the different ways we can use LLMs to generate patches: 1) generate the entire patch function, 2) fill in a chunk of code given the prefix and suffix 3) output a single line fix. We apply the LLMs under these repair settings on 5 datasets across 3 different languages and compare different LLMs in the number of bugs fixed, generation speed and compilation rate. We also compare the LLMs against recent state-of-the-art APR tools. Our study demonstrates that directly applying state-of-the-art LLMs can already substantially outperform all existing APR techniques on all our datasets. Among the studied LLMs, the scaling effect exists for APR where larger models tend to achieve better performance. Also, we show for the first time that suffix code after the buggy line (adopted in infilling-style APR) is important in not only generating more fixes but more patches with higher compilation rate. Besides patch generation, the LLMs consider correct patches to be more natural than other ones, and can even be leveraged for effective patch ranking or patch correctness checking. Lastly, we show that LLM-based APR can be further substantially boosted via: 1) increasing the sample size, and 2) incorporating fix template information.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
A carefully-designed demonstration based on the paper's findings can lead to substantial improvements over widely-used demonstration construction methods.
“We also show that a carefully-designed demonstration based on our findings can lead to substantial improvements over widely-used demonstration construction methods”
From What Makes Good In-Context Demonstrations for Code Intelligence Tasks with LLMs? (2023) · cites this paper as [17]
Existing studies mainly construct demonstrations by randomly selecting and arranging the demonstration examples.
“Existing studies [17], [19] mainly construct demonstrations by randomly selecting and arranging the demonstration examples.”
From What Makes Good In-Context Demonstrations for Code Intelligence Tasks with LLMs? (2023) · cites this paper as [17]
AGENTLESS targets complex repository-level issues spanning multiple locations, unlike simplistic bugs studied in most prior work.
“Furthermore, different from the simplistic bugs studied in most prior work [94], AGENTLESS targets complex repository-level issues spanning multiple locations.”
From Agentless: Demystifying LLM-based Software Engineering Agents (2024) · cites this paper as [94]
LLM-based APR tools, which sample multiple candidate patches per bug, have been shown to be state-of-the-art due to the powerful coding capability of modern LLMs.
“Specifically, LLM-based APR tools, which sample multiple candidate patches per bug, have been shown to be the state-of-the-art due to the powerful coding capability of modern LLMs [94].”
From Agentless: Demystifying LLM-based Software Engineering Agents (2024) · cites this paper as [94]
Cites
BLEU: a method for automatic evaluation of machine translation
Papineni, K., Roukos, S., Ward, T., & Zhu, W.-J. (2001) · Proceedings of the 40th Annual Meeting on Association for Computational Linguistics - ACL '02 · doi:10.3115/1073083.1073135 · Cited by 21,344 · see citing works
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929 · see citing works
Lewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.703 · Cited by 8,821 · see citing works
Scaling Laws for Neural Language Models
Kaplan, J., McCandlish, S., Henighan, T., Brown, T. B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., & Amodei, D. (2020) · arXiv · Cited by 1,749 · cited 2× · see citing works
“Using this idea and the exponential growth in LLM size [40], impressive performance in many tasks can be achieved even without any finetuning [27].”
Wang, Y., Wang, W., Joty, S., & Hoi, S. C. H. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.685 · Cited by 1,495 · see citing works
Wang, Y., Wang, W., Joty, S., & Hoi, S. C. H. (2021) · arXiv · doi:10.48550/arxiv.2109.00859 · Cited by 1,495 · see citing works
Defects4J: a database of existing faults to enable controlled testing studies for Java programs
Just, R., Jalali, D., & Ernst, M. D. (2014) · Proceedings of the 2014 International Symposium on Software Testing and Analysis · doi:10.1145/2610384.2628055 · Cited by 1,447 · see citing works
Evaluating Large Language Models Trained on Code
Chen, M., Tworek, J., Jun, H., Yuan, Q., Pinto, H. P. D. O., Kaplan, J., Edwards, H., Burda, Y., Joseph, N., Brockman, G., et al. (2021) · arXiv · doi:10.48550/arxiv.2107.03374 · Cited by 1,436 · see citing works
The Curious Case of Neural Text Degeneration
Holtzman, A., Buys, J., Du, L., Forbes, M., & Choi, Y. (2019) · arXiv · doi:10.48550/arxiv.1904.09751 · Cited by 1,304 · see citing works
Competition-level code generation with AlphaCode
Li, Y., Choi, D., Chung, J., Kushman, N., Schrittwieser, J., Leblond, R., Eccles, T., Keeling, J., Gimeno, F., Dal Lago, A., et al. (2022) · Science · doi:10.1126/science.abq1158 · Cited by 1,043 · see citing works
Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm
Reynolds, L., & McDonell, K. (2021) · Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411763.3451760 · Cited by 952 · see citing works
Fraser, G., & Arcuri, A. (2013) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2012.14 · Cited by 586 · see citing works
Shaping program repair space with existing patches and similar code
Jiang, J., Xiong, Y., Zhang, H., Gao, Q., & Chen, X. (2018) · Proceedings of the 27th ACM SIGSOFT International Symposium on Software Testing and Analysis · doi:10.1145/3213846.3213871 · Cited by 329 · see citing works
The ManyBugs and IntroClass benchmarks for automated repair of C programs
Le Goues, C., Holtschulte, N., Smith, E. K., Brun, Y., Devanbu, P., Forrest, S., & Weimer, W. (2015) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2015.2454513 · Cited by 326 · cited 1× · see citing works
“3) ManyBugs [64]: A C repair dataset consisting of 185 bugs gathered from 9 open-source projects with developer written tests.”
Toward automatic program synthesis
Manna, Z., & Waldinger, R. J. (1971) · Communications of the ACM · doi:10.1145/362566.362568 · Cited by 285 · see citing works
InCoder: A Generative Model for Code Infilling and Synthesis
Fried, D., Aghajanyan, A., Lin, J., Wang, S., Wallace, E., Shi, F., Zhong, R., Yih, W.-T., Zettlemoyer, L., & Lewis, M. (2022) · arXiv · doi:10.48550/arxiv.2204.05999 · Cited by 187 · cited 4× · see citing works
“• INCODER [33] A model designed for code infilling by adopting a causal masking objective [45].”
Influential
Flagged first, then by citations
Cited by
Sorted by citations
A survey on large language model (LLM) security and privacy: The Good, The Bad, and The Ugly
Yao, Y., Duan, J., Xu, K., Cai, Y., Sun, Z., & Zhang, Y. (2024) · High-Confidence Computing · doi:10.1016/j.hcc.2024.100211 · Cited by 1,083 · see citing works
Large Language Models for Software Engineering: A Systematic Literature Review
Hou, X., Zhao, Y., Liu, Y., Yang, Z., Wang, K., Li, L., Luo, X., Lo, D., Grundy, J., & Wang, H. (2024) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3695988 · Cited by 1,029 · see citing works
Software Testing With Large Language Models: Survey, Landscape, and Vision
Wang, J., Huang, Y., Chen, C., Liu, Z., Wang, S., & Wang, Q. (2024) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2024.3368208 · Cited by 573 · see citing works
AutoCodeRover: Autonomous Program Improvement
Zhang, Y., Ruan, H., Fan, Z., & Roychoudhury, A. (2024) · Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis · doi:10.1145/3650212.3680384 · Cited by 202 · see citing works
AutoCodeRover: Autonomous Program Improvement
Zhang, Y., Ruan, H., Fan, Z., & Roychoudhury, A. (2024) · arXiv · doi:10.48550/arxiv.2404.05427 · Cited by 202 · see citing works
Refining ChatGPT-Generated Code: Characterizing and Mitigating Code Quality Issues
Liu, Y., Le-Cong, T., Widyasari, R., Tantithamthavorn, C., Li, L., Le, X.-B. D., & Lo, D. (2024) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3643674 · Cited by 149 · see citing works
Wei, Y., Xia, C. S., & Zhang, L. (2023) · Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering · doi:10.1145/3611643.3616271 · Cited by 115 · see citing works
Self-citationWhat Makes Good In-Context Demonstrations for Code Intelligence Tasks with LLMs?
Gao, S., Wen, X.-C., Gao, C., Wang, W., Zhang, H., & Lyu, M. R. (2023) · 2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1109/ase56229.2023.00109 · Cited by 106 · see citing works
CodePlan: Repository-level Coding using LLMs and Planning
Bairi, R., Sonwane, A., Kanade, A., C, V. D., Iyer, A., Parthasarathy, S., Rajamani, S., Ashok, B., & Shet, S. (2023) · arXiv · doi:10.48550/arxiv.2309.12499 · Cited by 92 · see citing works
CodePlan: Repository-Level Coding using LLMs and Planning
Bairi, R., Sonwane, A., Kanade, A., C., V. D., Iyer, A., Parthasarathy, S., Rajamani, S., Ashok, B., & Shet, S. (2024) · Proceedings of the ACM on Software Engineering · doi:10.1145/3643757 · Cited by 92 · see citing works
Demystifying LLM-Based Software Engineering Agents
Xia, C. S., Deng, Y., Dunn, S., & Zhang, L. (2025) · Proceedings of the ACM on Software Engineering · doi:10.1145/3715754 · Cited by 87 · see citing works
Self-citationGenerative AI for self-adaptive systems: State of the art and research roadmap
Li, J., Zhang, M., Li, N., Weyns, D., Jin, Z., & Tei, K. (2024) · ACM Transactions on Autonomous and Adaptive Systems · doi:10.1145/3686803 · Cited by 77 · see citing works
Generative Artificial Intelligence for Software Engineering—A Research Agenda
Nguyen‐Duc, A., Cabrero‐Daniel, B., Przybylek, A., Arora, C., Khanna, D., Herda, T., Rafiq, U., Melegati, J., Guerra, E., Kemell, K., et al. (2025) · Software: Practice and Experience · doi:10.1002/spe.70005 · Cited by 60 · see citing works
The future of AI-driven software engineering
Terragni, V., Vella, A., Roop, P., & Blincoe, K. (2025) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3715003 · Cited by 62 · see citing works
Enhanced automated code vulnerability repair using large language models
de-Fitero-Dominguez, D., Garcia-Lopez, E., Garcia-Cabot, A., & Martinez-Herraiz, J.-J. (2024) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2024.109291 · Cited by 43 · see citing works
Shi, J., Yang, Z., & Lo, D. (2025) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3708525 · Cited by 43 · see citing works
CREF: An LLM-Based Conversational Software Repair Framework for Programming Tutors
Yang, B., Tian, H., Pian, W., Yu, H., Wang, H., Klein, J., Bissyandé, T. F., & Jin, S. (2024) · Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis · doi:10.1145/3650212.3680328 · Cited by 41 · see citing works
Agentless: Demystifying LLM-based Software Engineering Agents
Xia, C. S., Deng, Y., Dunn, S., & Zhang, L. (2024) · arXiv · doi:10.48550/arxiv.2407.01489 · Cited by 24 · see citing works
“We refer to this as the repository structure format , which begins with the root folder of the repository and organizes code files or folder names.”
MethodSelf-citationA 2030 Roadmap for Software Engineering
Pezzè, M., Abrahão, S., Penzenstadler, B., Poshyvanyk, D., Roychoudhury, A., & Yue, T. (2025) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3731559 · Cited by 13 · see citing works
Code Repair with LLMs gives an Exploration-Exploitation Tradeoff
Tang, H., Hu, K., Zhou, J. P., Zhong, S., Zheng, W.-L., Si, X., & Ellis, K. (2024) · arXiv · doi:10.48550/arxiv.2405.17503 · Cited by 11 · see citing works
Zhou, X., Weyssow, M., Widyasari, R., Zhang, T., He, J., Lyu, Y., Chang, J., Zhang, B., Huang, D., & Lo, D. (2025) · arXiv · doi:10.48550/arxiv.2502.06215 · Cited by 3 · see citing works
CodeCleaner: Mitigating Data Contamination for LLM Benchmarking
Cao, J., Chen, S., Zhang, W., Lo, H. C., Li, Y., & Cheung, S.-C. (2025) · Proceedings of the 16th International Conference on Internetware · doi:10.1145/3755881.3755901 · Cited by 2 · see citing works
Demonstration Attack against In-Context Learning for Code Intelligence
Ge, Y., Sun, W., Lou, Y., Fang, C., Zhang, Y., Li, Y., Zhang, X., Liu, Y., Zhao, Z., & Chen, Z. (2024) · arXiv · doi:10.48550/arxiv.2410.02841
“For example, Xia et al. [47] apply ICL to automated program repair tasks, leveraging demonstrations that guide the LLM in fixing errors in code, significantly improving the LLM’s repair capabilities and efficiency.”
Method
- Authors
- Xia, C. S., Wei, Y., & Zhang, L.
- Affiliations
- University of Illinois Urbana-Champaign (US)
- Venue
- 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE)
- Publisher
- IEEE
- Published
- 2023-05
- Pages
- 1482-1494
- ISSN
- 0270-5257 (ISSN-L)
- OpenAlex
- W4384345708
- Cited by
- 433
- References
- 16
- Citations by year
- 2026: 105 · 2025: 157 · 2024: 141 · 2023: 30
- FWCI
- 83.75
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM/IEEE International Conference on Software Engineering: Companion Proceedings (ICSE-Companion) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- NSF (CCF-2131943,CCF-2141474)
- Topics
- Physical Sciences › Computer Science › Software › Software Testing and Debugging Techniques
- Physical Sciences › Computer Science › Information Systems › Software Engineering Research
- Physical Sciences › Computer Science › Computer Networks and Communications › Software System Performance and Reliability
- Keywords
- automated program repair, large pre-trained language models, bug fixes, patch generation, infilling, model scaling
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
