---
title: "Kimi-Dev: Agentless Training as Skill Prior for SWE-Agents (2025)"
description: "Automated GitHub issue fixing, as measured by SWE-bench, has split into two camps. 50 references, cited by 4 papers here."
authors: ["Zonghan Yang", "Shengjie Wang", "Kelin Fu", "Wenyang He", "Weimin Xiong", "Yibo Liu", "Yibo Miao", "Bofei Gao", "Yejie Wang", "Yingwei Ma", "Yanhao Li", "Yue Liu", "Zhenxing Hu", "Kaitai Zhang", "Shuyi Wang", "Huarong Chen", "Flood Sung", "Yang Liu", "Yang Gao", "Zhilin Yang", "Tianyu Liu"]
year: 2025
venue: "arXiv:2509.23045"
source: "https://arxiv.org/abs/2509.23045"
doi: "10.48550/arxiv.2509.23045"
arxiv: "2509.23045"
licence: "CC BY 4.0 (arXiv)"
licence_url: "http://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/yang-2025-kimi-dev-agentless-training/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/yang-2025-kimi-dev-agentless-training/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2025](https://md.research.elamur.ai/free/en/papers/2025/page/11/) / Kimi-Dev: Agentless Training as Skill Prior for SWE-Agents

# Kimi-Dev: Agentless Training as Skill Prior for SWE-Agents

**Authors:** [Zonghan Yang](https://md.research.elamur.ai/free/en/authors/zonghan-yang/), Shengjie Wang, Kelin Fu, Wenyang He, [Weimin Xiong](https://md.research.elamur.ai/free/en/authors/weimin-xiong/), Yibo Liu, [Yibo Miao](https://md.research.elamur.ai/free/en/authors/yibo-miao/), [Bofei Gao](https://md.research.elamur.ai/free/en/authors/bofei-gao/), Yejie Wang, [Yingwei Ma](https://md.research.elamur.ai/free/en/authors/yingwei-ma/), Yanhao Li, [Yue Liu](https://md.research.elamur.ai/free/en/authors/yue-liu/), Zhenxing Hu, Kaitai Zhang, Shuyi Wang, [Huarong Chen](https://md.research.elamur.ai/free/en/authors/huarong-chen/), Flood Sung, [Yang Liu](https://md.research.elamur.ai/free/en/authors/yang-liu/), [Yang Gao](https://md.research.elamur.ai/free/en/authors/yang-gao/), [Zhilin Yang](https://md.research.elamur.ai/free/en/authors/zhilin-yang/) and [Tianyu Liu](https://md.research.elamur.ai/free/en/authors/tianyu-liu/) · **Year:** 2025 · **Venue:** arXiv:2509.23045 · **Licence:** [CC BY 4.0 (arXiv)](http://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2509.23045) · [PDF](https://arxiv.org/pdf/2509.23045) · [DOI 10.48550/arxiv.2509.23045](https://doi.org/10.48550/arxiv.2509.23045) · [arXiv 2509.23045](https://arxiv.org/abs/2509.23045) · [OpenAlex](https://openalex.org/W4415332089)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · [4](https://md.research.elamur.ai/free/en/papers/yang-2025-kimi-dev-agentless-training/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary

Automated GitHub issue fixing, as measured by SWE-bench, has split into two camps. Multi-turn SWE-Agents are flexible but hard to train end-to-end. Pipeline-style Agentless methods are easy to optimize with verifiable rewards but rigid. The authors argue the two approaches can be combined: training on Agentless workflows instills reusable "skill priors" such as file localization, code editing, and self-reflection, which then help a model adapt to agentic use.

Starting from Qwen 2.5-72B-Base, they build Kimi-Dev in four stages:
- large-scale mid-training on GitHub issue and PR data,
- a reasoning cold-start using DeepSeek R1 trajectories,
- reinforcement learning,
- test-time self-play between a BugFixer role and a TestWriter role.

The resulting open model reaches 60.4% on SWE-bench Verified, the best reported score for workflow-based systems. A brief SFT on about 5k public agent trajectories then lifts it to 48.6% pass@1 as an SWE-Agent. Priors with Agentless training also generalize better to other benchmarks, especially when adaptation data is scarce. This points to a cheaper and more stable path to open coding agents.

## Key points

- **Mid-training data (~150B tokens):** about 50B tokens of Agentless-formatted data derived from diff patches, about 20B of curated PR commit packs, and about 20B of synthetic reasoning and agentic data (upsampled 4×). Repositories from the SWE-bench Verified test set were excluded to prevent contamination.
- **Cold-start data:** built from SWE-Gym and SWE-bench-extra, with DeepSeek R1 (20,250,120) acting as both BugFixer and TestWriter.
- **Agentless result:** Kimi-Dev (72B) scores 60.4% on SWE-bench Verified. Weights are released as `moonshotai/Kimi-Dev-72B` on HuggingFace, with code on GitHub (MoonshotAI/Kimi-Dev).
- **Agentic result:** SFT on roughly 5,016 SWE-smith trajectories gives 48.6% pass@1, comparable to Claude 3.5 Sonnet (20,241,022). Each adaptation run takes from a few minutes to about two hours.
- **Generalization tests:**
  - SWE-bench-live: 300 tasks collected October 2024 to March 2025.
  - SWE-bench Multilingual: 300 tasks from 42 repositories in 9 languages.
  - The SFT and RL priors beat the Base prior in low-data settings, but the gap narrows as more trajectories are used.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

📚 Shelves (3): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🗄️ Information Systems](https://md.research.elamur.ai/free/en/shelves/computer-science/information-systems/) · 1,437 works · [💻 Software](https://md.research.elamur.ai/free/en/shelves/computer-science/software/) · 92 works

## Cite

Cite as: Zonghan Yang, Shengjie Wang, Kelin Fu, Wenyang He, Weimin Xiong, Yibo Liu, Yibo Miao, Bofei Gao, Yejie Wang, Yingwei Ma, Yanhao Li, Yue Liu, Zhenxing Hu, Kaitai Zhang, Shuyi Wang, Huarong Chen, Flood Sung, Yang Liu, Yang Gao, Zhilin Yang and Tianyu Liu. 2025. Kimi-Dev: Agentless Training as Skill Prior for SWE-Agents. arXiv:2509.23045.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@misc{yang2025kimi,
  title = {{Kimi-Dev: Agentless Training as Skill Prior for SWE-Agents}},
  author = {Zonghan Yang, Shengjie Wang, Kelin Fu, Wenyang He, Weimin Xiong, Yibo Liu, Yibo Miao, Bofei Gao, Yejie Wang, Yingwei Ma, Yanhao Li, Yue Liu, Zhenxing Hu, Kaitai Zhang, Shuyi Wang, Huarong Chen, Flood Sung, Yang Liu, Yang Gao, Zhilin Yang and Tianyu Liu},
  year = {2025},
  doi = {10.48550/arxiv.2509.23045},
  eprint = {2509.23045},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2509.23045}
}
```

- [← previous In the LLM era, Word Sense Induction remains unsolved](https://md.research.elamur.ai/free/en/papers/mosolova-2025-wsi-unsolved/)
- [next → Kuwain 1.5B: An Arabic SLM via Language Injection](https://md.research.elamur.ai/free/en/papers/hennara-2025-kuwain-1-5b-arabic-slm/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/11/)

## References

50 references, 20 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Mistral AI and All Hands AI. devstral-2507, July 2025. [find](https://scholar.google.com/scholar?q=Mistral+AI+and+All+Hands+AI%2E+devstral%2D2507%2C+July+2025%2E)
2. Devstral-small-2507. [https://mistral.ai/news/Anthropic](https://mistral.ai/news/Anthropic). Raising the bar on swe-bench verified with claude 3.5 sonnet. Online; AI model, Oct 2024. URL https://www.anthropic.com/engineering/swe-bench-sonnet. [↗](https://mistral.ai/news/Anthropic)
3. Anthropic. Claude 3.7 sonnet: Hybrid reasoning model. [https://www.anthropic.com/](https://www.anthropic.com/) news/claude-3-7-sonnet, February 2025a. [↗](https://www.anthropic.com/)
4. Claude sonnet 4. [https://www.anthropic.com/news/claude-4](https://www.anthropic.com/news/claude-4), May Ibragim Badertdinov, Maria Trofimova, Yuri Anapolskiy, Sergey Abramov, Karina Zainullina, Alexander Golubev, Sergey Polezhaev, Daria Litvintseva, Simon Karasik, Filipp Fisin, et al. Scaling data collection for training software engineerin… [↗](https://www.anthropic.com/news/claude-4)
5. Ibragim Badertdinov, Maria Trofimova, Yury Anapolskiy, Sergey Abramov, Karina Zainullina, Alexander Golubev, Sergey Polezhaev, Daria Litvintseva, Simon Karasik, Filipp Fisin, Sergey Skvortsov, Maxim Nekrashevich, Anton Shevtsov, and Boris Yangel. Scaling data collection for training software enginee… [find](https://scholar.google.com/scholar?q=Ibragim+Badertdinov%2C+Maria+Trofimova%2C+Yury+Anapolskiy%2C+Sergey+Abramov%2C+Karina+Zainullina%2C+Alexander+Golubev%2C+Sergey+Polezhaev%2C+Daria+Litvintseva%2C+Simon+Karasik%2C+Filipp+Fisin%2C+Sergey+Skvortsov%2C+Maxim+N)
6. Ibragim Badertdinov, Alexander Golubev, Maksim Nekrashevich, Anton Shevtsov, Simon Karasik, Andrei Andriushchenko, Maria Trofimova, Daria Litvintseva, and Boris Yangel. Swe-rebench: An automated pipeline for task collection and decontaminated evaluation of software engineering agents, 2025. URL http… — [arXiv:2505.20411](https://arxiv.org/abs/2505.20411) [find](https://scholar.google.com/scholar?q=Ibragim+Badertdinov%2C+Alexander+Golubev%2C+Maksim+Nekrashevich%2C+Anton+Shevtsov%2C+Simon+Karasik%2C+Andrei+Andriushchenko%2C+Maria+Trofimova%2C+Daria+Litvintseva%2C+and+Boris+Yangel%2E+Swe%2Drebench%3A+An+automated+pipel)
7. Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter,… — [Language Models are Few-Shot Learners](https://md.research.elamur.ai/free/en/papers/brown-2020-language-models-few-shot-learners/)
8. Brendan Burns, Brian Grant, David Oppenheimer, Eric Brewer, and John Wilkes. Borg, Omega, and Kubernetes. In Proceedings of the 10th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’16), pp. 285–301, 2016. [find](https://scholar.google.com/scholar?q=Brendan+Burns%2C+Brian+Grant%2C+David+Oppenheimer%2C+Eric+Brewer%2C+and+John+Wilkes%2E+Borg%2C+Omega%2C+and+Kubernetes%2E+In+Proceedings+of+the+10th+USENIX+Symposium+on+Operating+Systems+Design+and+Implementation+%28OS)
9. Shiyi Cao, Sumanth Hegde, Dacheng Li, Tyler Griggs, Shu Liu, Eric Tang, Jiayi Pan, Xingyao Wang, Akshay Malik, Graham Neubig, Kourosh Hakhamaneshi, Richard Liaw, Philipp Moritz, Matei Zaharia, Joseph E. Gonzalez, and Ion Stoica. Skyrl-v0: Train real-world long-horizon agents via reinforcement learni… [find](https://scholar.google.com/scholar?q=Shiyi+Cao%2C+Sumanth+Hegde%2C+Dacheng+Li%2C+Tyler+Griggs%2C+Shu+Liu%2C+Eric+Tang%2C+Jiayi+Pan%2C+Xingyao+Wang%2C+Akshay+Malik%2C+Graham+Neubig%2C+Kourosh+Hakhamaneshi%2C+Richard+Liaw%2C+Philipp+Moritz%2C+Matei+Zaharia%2C+Joseph+)
10. Aili Chen, Aonian Li, Bangwei Gong, Binyang Jiang, Bo Fei, Bo Yang, Boji Shan, Changqing Yu, Chao Wang, Cheng Zhu, et al. [Minimax-m1: Scaling test-time compute efficiently with lightning attention](https://md.research.elamur.ai/free/en/papers/minimax-2025-minimax-m1-scaling-test/). arXiv preprint arXiv:2506.13585, 2025a.
11. Mouxiang Chen, Binyuan Hui, Zeyu Cui, Jiaxi Yang, Dayiheng Liu, Jianling Sun, Junyang Lin, and Zhongxin Liu. Parallel scaling law for language models. arXiv preprint arXiv:2505.10475, 2025b. — [arXiv:2505.10475](https://arxiv.org/abs/2505.10475) [find](https://scholar.google.com/scholar?q=Mouxiang+Chen%2C+Binyuan+Hui%2C+Zeyu+Cui%2C+Jiaxi+Yang%2C+Dayiheng+Liu%2C+Jianling+Sun%2C+Junyang+Lin%2C+and+Zhongxin+Liu%2E+Parallel+scaling+law+for+language+models%2E+arXiv+preprint+arXiv%3A2505%2E10475%2C+2025b%2E+%E2%80%94+%5BarXiv%3A)
12. Tianzhe Chu, Yuexiang Zhai, Jihan Yang, Shengbang Tong, Saining Xie, Dale Schuurmans, Quoc V Le, Sergey Levine, and Yi Ma. [SFT memorizes, RL generalizes: A comparative study of foundation model post-training](https://md.research.elamur.ai/free/en/papers/chu-2025-sft-memorizes-rl-generalizes-comparative/). In Forty-second International Conference on Machine Learning, 2025. URL [https://openreview.…](https://openreview.…) [↗](https://openreview.…)
13. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, et al. [Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities](https://md.research.elamur.ai/free/en/papers/comanici-2025-gemini-2-5-pushing-frontier/). arXiv preprint a…
14. Chelsea Finn, Pieter Abbeel, and Sergey Levine. [Model-agnostic meta-learning for fast adaptation of deep networks](https://md.research.elamur.ai/free/en/papers/finn-2017-model-agnostic-meta-learning-fast/). In International conference on machine learning, pp. 1126–1135. PMLR, 2017.
15. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. [The llama 3 herd of models](https://md.research.elamur.ai/free/en/papers/grattafiori-2024-llama-3-herd-models/). arXiv preprint arXiv:2407.21783, 2024.
16. Xinran Gu, Kaifeng Lyu, Jiazheng Li, and Jingzhao Zhang. Data mixing can induce phase transitions in knowledge acquisition. arXiv preprint arXiv:2505.18091, 2025. — [arXiv:2505.18091](https://arxiv.org/abs/2505.18091) [find](https://scholar.google.com/scholar?q=Xinran+Gu%2C+Kaifeng+Lyu%2C+Jiazheng+Li%2C+and+Jingzhao+Zhang%2E+Data+mixing+can+induce+phase+transitions+in+knowledge+acquisition%2E+arXiv+preprint+arXiv%3A2505%2E18091%2C+2025%2E+%E2%80%94+%5BarXiv%3A2505%2E18091%5D%28https%3A%2F%2Farxiv%2Eor)
17. Arnav Gudibande, Eric Wallace, Charlie Victor Snell, Xinyang Geng, Hao Liu, Pieter Abbeel, Sergey Levine, and Dawn Song. The False Promise of Imitating Proprietary Language Models. In The Twelfth International Conference on Learning Representations, 2024. URL [https://openreview.net/forum?id=Kz3yckp…](https://openreview.net/forum?id=Kz3yckp…) [↗](https://openreview.net/forum?id=Kz3yckp…)
18. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. [Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning](https://md.research.elamur.ai/free/en/papers/deepseek-ai-2025-deepseek-r1-incentivizing-reasoning-capability/). arXiv preprint arXiv:2501.12948, 2025.
19. Zhenyu He, Qingping Yang, Wei Sheng, Xiaojian Zhong, Kechi Zhang, Chenxin An, Wenlei Shi, Tianle Cai, Di He, Jiaze Chen, Jingjing Xu, and Mingxuan Wang. Swe-swiss: A multi-task fine-tuning and rl recipe for high-performance issue resolution. Notion page, 2025. [find](https://scholar.google.com/scholar?q=Zhenyu+He%2C+Qingping+Yang%2C+Wei+Sheng%2C+Xiaojian+Zhong%2C+Kechi+Zhang%2C+Chenxin+An%2C+Wenlei+Shi%2C+Tianle+Cai%2C+Di+He%2C+Jiaze+Chen%2C+Jingjing+Xu%2C+and+Mingxuan+Wang%2E+Swe%2Dswiss%3A+A+multi%2Dtask+fine%2Dtuning+and+rl+reci)
20. Naman Jain, Jaskirat Singh, Manish Shetty, Liang Zheng, Koushik Sen, and Ion Stoica. R2e-gym: Procedural environments and hybrid verifiers for scaling open-weights swe agents. arXiv preprint arXiv:2504.07164, 2025. — [arXiv:2504.07164](https://arxiv.org/abs/2504.07164) [find](https://scholar.google.com/scholar?q=Naman+Jain%2C+Jaskirat+Singh%2C+Manish+Shetty%2C+Liang+Zheng%2C+Koushik+Sen%2C+and+Ion+Stoica%2E+R2e%2Dgym%3A+Procedural+environments+and+hybrid+verifiers+for+scaling+open%2Dweights+swe+agents%2E+arXiv+preprint+arXiv%3A250)
21. Carlos E Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, and Karthik Narasimhan. [Swe-bench: Can language models resolve real-world github issues?](https://md.research.elamur.ai/free/en/papers/jimenez-2024-swe-bench-can-language-models/) arXiv preprint arXiv:2310.06770, 2023.
22. Team Kimi, Yifan Bai, Yiping Bao, Guanduo Chen, Jiahao Chen, Ningxin Chen, Ruijue Chen, Yanru Chen, Yuankun Chen, Yutian Chen, et al. [Kimi k2: Open agentic intelligence](https://md.research.elamur.ai/free/en/papers/kimi-team-2025-kimi-k2-open-agentic/). arXiv preprint arXiv:2507.20534, 2025.
23. Michael Luo, Naman Jain, Jaskirat Singh, Sijun Tan, Ameen Patel, Qingyang Wu, Alpay Ariyak, Colin Cai, Shang Zhu Tarun Venkat, Ben Athiwaratkun, Manan Roongta, Ce Zhang, Li Erran Li, Raluca Ada Popa, Koushik Sen, and Ion Stoica. DeepSWE: Training a State-of-the-Art Coding Agent from Scratch by Scali… [find](https://scholar.google.com/scholar?q=Michael+Luo%2C+Naman+Jain%2C+Jaskirat+Singh%2C+Sijun+Tan%2C+Ameen+Patel%2C+Qingyang+Wu%2C+Alpay+Ariyak%2C+Colin+Cai%2C+Shang+Zhu+Tarun+Venkat%2C+Ben+Athiwaratkun%2C+Manan+Roongta%2C+Ce+Zhang%2C+Li+Erran+Li%2C+Raluca+Ada+Popa%2C+)
24. Yingwei Ma, Rongyu Cao, Yongchang Cao, Yue Zhang, Jue Chen, Yibo Liu, Yuchen Liu, Binhua Li, Fei Huang, and Yongbin Li. Swe-gpt: A process-centric language model for automated software improvement. Proc. ACM Softw. Eng., 2(ISSTA), June 2025a. doi: 10.1145/3728981. URL https://doi.org/10.1145/3728981… — [doi:10.1145/3728981](https://doi.org/10.1145/3728981) [↗](https://doi.org/10.1145/3728981…)
25. Yingwei Ma, Qingping Yang, Rongyu Cao, Binhua Li, Fei Huang, and Yongbin Li. Alibaba ling-maagent: Improving automated issue resolution via comprehensive repository exploration. In Proceedings of the 33rd ACM International Conference on the Foundations of Software Engineering, pp. 238–249, 2025b. [find](https://scholar.google.com/scholar?q=Yingwei+Ma%2C+Qingping+Yang%2C+Rongyu+Cao%2C+Binhua+Li%2C+Fei+Huang%2C+and+Yongbin+Li%2E+Alibaba+ling%2Dmaagent%3A+Improving+automated+issue+resolution+via+comprehensive+repository+exploration%2E+In+Proceedings+of+the+)
26. OpenAI. [Openai o1 system card](https://md.research.elamur.ai/free/en/papers/openai-2024-openai-o1-system-card/). Technical report, OpenAI, 2024. URL [https://arxiv.org/](https://arxiv.org/) abs/2412.16720. Includes o1 and o1-mini models, safety and evaluation work. [↗](https://arxiv.org/)
27. OpenAI. Introducing codex. [https://openai.com/index/introducing-codex/](https://openai.com/index/introducing-codex/), May 2025. [↗](https://openai.com/index/introducing-codex/)
28. OpenAI. Openai o3 and openai o4-mini system card. Technical report, OpenAI, Apr 2025. URL [https://cdn.openai.com/pdf/2221c875-02dc-4789-800b-e7758f3722c1/](https://cdn.openai.com/pdf/2221c875-02dc-4789-800b-e7758f3722c1/) o3-and-o4-mini-system-card.pdf. [↗](https://cdn.openai.com/pdf/2221c875-02dc-4789-800b-e7758f3722c1/)
29. OpenAI. Openai o3-mini system card. Technical report / system card, OpenAI, 2025a. URL [https://cdn.openai.com/o3-mini-system-card-feb10.pdf](https://cdn.openai.com/o3-mini-system-card-feb10.pdf). Description of o3-mini model, its safety evaluations and reasoning benchmarks. [↗](https://cdn.openai.com/o3-mini-system-card-feb10.pdf)
30. OpenAI. [gpt-oss-120b & gpt-oss-20b model card](https://md.research.elamur.ai/free/en/papers/agarwal-2025-gpt-oss-120b-gpt-oss/), 2025b. URL https://arxiv.org/abs/ 2508.10925. [↗](https://arxiv.org/abs/)
31. OpenAI. Gpt-5 system card. OpenAI website, Aug 2025c. URL — [OpenAI GPT-5 System Card](https://md.research.elamur.ai/free/en/papers/singh-2025-openai-gpt-5-system-card/) [https://cdn.openai.com/](https://cdn.openai.com/) gpt-5-system-card.pdf. [↗](https://cdn.openai.com/)
32. Jiayi Pan, Xingyao Wang, Graham Neubig, Navdeep Jaitly, Heng Ji, Alane Suhr, and Yizhe Zhang. Training software engineering agents and verifiers with swe-gym. arXiv preprint arXiv:2412.21139, 2024. [find](https://scholar.google.com/scholar?q=Jiayi+Pan%2C+Xingyao+Wang%2C+Graham+Neubig%2C+Navdeep+Jaitly%2C+Heng+Ji%2C+Alane+Suhr%2C+and+Yizhe+Zhang%2E+Training+software+engineering+agents+and+verifiers+with+swe%2Dgym%2E+arXiv+preprint+arXiv%3A2412%2E21139%2C+2024%2E)
33. Jiayi Pan, Xiuyu Li, Long Lian, Charlie Victor Snell, Yifei Zhou, Adam Yala, Trevor Darrell, Kurt Keutzer, and Alane Suhr. Learning adaptive parallel reasoning with language models. In Second Conference on Language Modeling, 2025. URL [https://openreview.net/forum?id=](https://openreview.net/forum?id=) YgwQ7sXPXU. [↗](https://openreview.net/forum?id=)
34. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Mei Li, Mingfeng… [find](https://scholar.google.com/scholar?q=Qwen%2C+%3A%2C+An+Yang%2C+Baosong+Yang%2C+Beichen+Zhang%2C+Binyuan+Hui%2C+Bo+Zheng%2C+Bowen+Yu%2C+Chengyuan+Li%2C+Dayiheng+Liu%2C+Fei+Huang%2C+Haoran+Wei%2C+Huan+Lin%2C+Jian+Yang%2C+Jianhong+Tu%2C+Jianwei+Zhang%2C+Jianxin+Yang%2C+Jiaxi+)
35. ByteDance Seed, Jiaze Chen, Tiantian Fan, Xin Liu, Lingjun Liu, Zhiqi Lin, Mingxuan Wang, Chengyi Wang, Xiangpeng Wei, Wenyuan Xu, et al. Seed1. 5-thinking: Advancing superb reasoning models with reinforcement learning. arXiv preprint arXiv:2504.[13914, 2025](https://md.research.elamur.ai/free/en/papers/bytedance-seed-2025-seed1-5-thinking-advancing/).
36. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, et al. [Deepseekmath: Pushing the limits of mathematical reasoning in open language models](https://md.research.elamur.ai/free/en/papers/shao-2024-deepseekmath-pushing-limits-mathematical-reasoning/). arXiv preprint arXiv:2402.03300, 2024.
37. Kimi Team, Angang Du, Bofei Gao, Bowei Xing, Changjiu Jiang, Cheng Chen, Cheng Li, Chenjun Xiao, Chenzhuang Du, Chonghua Liao, et al. Kimi k1. 5: Scaling reinforcement learning with llms. arXiv preprint arXiv:2501.[12599, 2025](https://md.research.elamur.ai/free/en/papers/kimi-team-2025-kimi-k1-5-scaling-reinforcement/).
38. Xingyao Wang, Boxuan Li, Yufan Song, Frank F. Xu, Xiangru Tang, Mingchen Zhuge, Jiayi Pan, Yueqi Song, Bowen Li, Jaskirat Singh, Hoang H. Tran, Fuqiang Li, Ren Ma, Mingzhang Zheng, Bill Qian, Yanjun Shao, Niklas Muennighoff, Yizhe Zhang, Binyuan Hui, Junyang Lin, Robert Brennan, Hao Peng, Heng Ji, a… — [OpenHands: An Open Platform for AI Software Developers as Generalist Agents](https://md.research.elamur.ai/free/en/papers/wang-2024-openhands-open-platform-ai/)
39. Xingyao Wang, Boxuan Li, Yufan Song, Frank F. Xu, Xiangru Tang, Mingchen Zhuge, Jiayi Pan, Yueqi Song, Bowen Li, Jaskirat Singh, Hoang H. Tran, Fuqiang Li, Ren Ma, Mingzhang Zheng, Bill Qian, Yanjun Shao, Niklas Muennighoff, Yizhe Zhang, Binyuan Hui, Junyang Lin, Robert Brennan, Hao Peng, Heng Ji, a… — [OpenHands: An Open Platform for AI Software Developers as Generalist Agents](https://md.research.elamur.ai/free/en/papers/wang-2024-openhands-open-platform-ai/)
40. Yuxiang Wei, Olivier Duchenne, Jade Copet, Quentin Carbonneaux, Lingming Zhang, Daniel Fried, Gabriel Synnaeve, Rishabh Singh, and Sida I Wang. Swe-rl: Advancing llm reasoning via reinforcement learning on open software evolution. arXiv preprint arXiv:2502.18449, 2025. [find](https://scholar.google.com/scholar?q=Yuxiang+Wei%2C+Olivier+Duchenne%2C+Jade+Copet%2C+Quentin+Carbonneaux%2C+Lingming+Zhang%2C+Daniel+Fried%2C+Gabriel+Synnaeve%2C+Rishabh+Singh%2C+and+Sida+I+Wang%2E+Swe%2Drl%3A+Advancing+llm+reasoning+via+reinforcement+learni)
41. Ronald J Williams. Simple statistical gradient-following algorithms for connectionist reinforcement learning. Machine learning, 8(3):229–256, 1992. [find](https://scholar.google.com/scholar?q=Ronald+J+Williams%2E+Simple+statistical+gradient%2Dfollowing+algorithms+for+connectionist+reinforcement+learning%2E+Machine+learning%2C+8%283%29%3A229%E2%80%93256%2C+1992%2E)
42. Chunqiu Steven Xia, Yinlin Deng, Soren Dunn, and Lingming Zhang. Agentless: Demystifying llm-based software engineering agents. arXiv preprint arXiv:2407.01489, 2024. [find](https://scholar.google.com/scholar?q=Chunqiu+Steven+Xia%2C+Yinlin+Deng%2C+Soren+Dunn%2C+and+Lingming+Zhang%2E+Agentless%3A+Demystifying+llm%2Dbased+software+engineering+agents%2E+arXiv+preprint+arXiv%3A2407%2E01489%2C+2024%2E)
43. Chengxing Xie, Bowen Li, Chang Gao, He Du, Wai Lam, Difan Zou, and Kai Chen. SWE-fixer: Training open-source LLMs for effective and efficient GitHub issue resolution. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar (eds.), Findings of the Association for Computational… [find](https://scholar.google.com/scholar?q=Chengxing+Xie%2C+Bowen+Li%2C+Chang+Gao%2C+He+Du%2C+Wai+Lam%2C+Difan+Zou%2C+and+Kai+Chen%2E+SWE%2Dfixer%3A+Training+open%2Dsource+LLMs+for+effective+and+efficient+GitHub+issue+resolution%2E+In+Wanxiang+Che%2C+Joyce+Nabende%2C+E)
44. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. [Qwen3 technical report](https://md.research.elamur.ai/free/en/papers/yang-2025-qwen3-technical-report/). arXiv preprint arXiv:2505.09388, 2025a.
45. John Yang, Carlos E Jimenez, Alexander Wettig, Kilian Lieret, Shunyu Yao, Karthik Narasimhan, and Ofir Press. [Swe-agent: Agent-computer interfaces enable automated software engineering](https://md.research.elamur.ai/free/en/papers/yang-2024-swe-agent-agent-computer-interfaces/). Advances in Neural Information Processing Systems, 37:50528–50652, 2024a.
46. John Yang, Kilian Leret, Carlos E Jimenez, Alexander Wettig, Kabir Khandpur, Yanzhe Zhang, Binyuan Hui, Ofir Press, Ludwig Schmidt, and Diyi Yang. Swe-smith: Scaling data for software engineering agents. arXiv preprint arXiv:2504.21798, 2025b. [find](https://scholar.google.com/scholar?q=John+Yang%2C+Kilian+Leret%2C+Carlos+E+Jimenez%2C+Alexander+Wettig%2C+Kabir+Khandpur%2C+Yanzhe+Zhang%2C+Binyuan+Hui%2C+Ofir+Press%2C+Ludwig+Schmidt%2C+and+Diyi+Yang%2E+Swe%2Dsmith%3A+Scaling+data+for+software+engineering+agen)
47. John Yang, Kilian Lieret, Carlos E. Jimenez, Alexander Wettig, Kabir Khandpur, Yanzhe Zhang, Binyuan Hui, Ofir Press, Ludwig Schmidt, and Diyi Yang. Swe-smith: Scaling data for software engineering agents, 2025c. URL https://arxiv.org/abs/2504.21798. [↗](https://arxiv.org/abs/2504.21798)
48. Zonghan Yang, Peng Li, Ming Yan, Ji Zhang, Fei Huang, and Yang Liu. ReAct Meets ActRe: Autonomous Annotation of Agent Trajectories for Contrastive Self-Training. In First Conference on Language Modeling, 2024b. URL [https://openreview.net/forum?id=](https://openreview.net/forum?id=) 0VLBwQGWpA. [↗](https://openreview.net/forum?id=)
49. Liang Zeng, Yongcong Li, Yuzhen Xiao, Changshi Li, Chris Yuhao Liu, Rui Yan, Tianwen Wei, Jujie He, Xuchen Song, Yang Liu, et al. [Skywork-swe: Unveiling data scaling laws for software engineering in llms](https://md.research.elamur.ai/free/en/papers/zeng-2025-skywork-swe-unveiling-data/). arXiv preprint arXiv:2506.19290, 2025.
50. Linghao Zhang, Shilin He, Chaoyun Zhang, Yu Kang, Bowen Li, Chengxing Xie, Junhao Wang, Maoquan Wang, Yufan Huang, Shengyu Fu, et al. Swe-bench goes live! arXiv preprint arXiv:2505.23419, 2025. — [arXiv:2505.23419](https://arxiv.org/abs/2505.23419) [find](https://scholar.google.com/scholar?q=Linghao+Zhang%2C+Shilin+He%2C+Chaoyun+Zhang%2C+Yu+Kang%2C+Bowen+Li%2C+Chengxing+Xie%2C+Junhao+Wang%2C+Maoquan+Wang%2C+Yufan+Huang%2C+Shengyu+Fu%2C+et+al%2E+Swe%2Dbench+goes+live%21+arXiv+preprint+arXiv%3A2505%2E23419%2C+2025%2E+%E2%80%94+%5BarX)

*Updated 31 October 2026*