用单轮代码训练构建技能先验，让代码代理更高效。
Kimi-Dev: Agentless Training as Skill Prior for SWE-Agents
- 通过单轮可验证步骤训练，学习定位、编辑和自省等编码技能
- 在SWE-bench上达到60.4%的验证通过率，是工作流方法最佳表现
- 适配后性能媲美Claude 3.5 Sonnet，适合想低成本构建代码代理的研究者
大语言模型在软件工程领域应用日益广泛，以SWE-bench为关键评估基准。现有方案分为支持多轮交互的SWE-Agent框架与仅需单轮验证的workflow-based Agentless方法。本文认为二者并非互斥：推理密集型的Agentless训练可生成技能先验，包括代码定位、编辑与自省能力，从而实现高效有效的SWE-Agent适应。本文首先构建了Agentless训练范式，提出开源模型Kimi-Dev，在SWE-bench Verified上达到60.4%的通过率，为当前工作流方法最优表现。经5千条公开轨迹的SFT微调后，其驱动的SWE-Agents在pass@1指标上达48.6%，与Claude 3.5 Sonnet（241022版本）相当。结果表明，来自Agentless训练的结构化技能先验可有效连接工作流与代理式框架，推动可迁移编码代理的发展。
原文摘要 · Abstract (English)
Large Language Models (LLMs) are increasingly applied to software engineering (SWE), with SWE-bench as a key benchmark. Solutions are split into SWE-Agent frameworks with multi-turn interactions and workflow-based Agentless methods with single-turn verifiable steps. We argue these paradigms are not mutually exclusive: reasoning-intensive Agentless training induces skill priors, including localization, code edit, and self-reflection that enable efficient and effective SWE-Agent adaptation. In this work, we first curate the Agentless training recipe and present Kimi-Dev, an open-source SWE LLM achieving 60.4\% on SWE-bench Verified, the best among workflow approaches. With additional SFT adaptation on 5k publicly-available trajectories, Kimi-Dev powers SWE-Agents to 48.6\% pass@1, on par with that of Claude 3.5 Sonnet (241022 version). These results show that structured skill priors from Agentless training can bridge workflow and agentic frameworks for transferable coding agents.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。