无需环境即可验证代码补丁，提升编程代理训练效率
Dockerless: Environment-Free Program Verifier for Coding Agents
- 通过智能探索代码库收集证据，判断补丁正确性，不依赖运行环境
- 在基准测试中比最强开源验证器高14.3 AUC，SWE-bench验证集解决率达62.0%
- 适用于无环境部署的编程模型训练，适合追求高效低耗的开发者
程序验证在训练编程代理中起核心作用，包括为监督微调（SFT）筛选轨迹和为强化学习（RL）提供奖励。传统基于执行的验证需在每个仓库的Docker镜像等环境中运行单元测试，带来高昂的环境配置成本。我们提出Dockerless，一种无需环境的代理式补丁验证方法，可在不执行代码的情况下评估生成的代码补丁。不同于简单匹配候选补丁与参考答案，Dockerless通过智能代理探索代码库收集证据来判断补丁正确性。在验证器评估基准上，Dockerless比最强的开源验证器高出14.3 AUC点。将其同时用于SFT轨迹过滤和RL奖励，可实现完全无环境的后训练流程。最终模型在SWE-bench Verified、Multilingual和Pro上的解决率分别为62.0%、50.0%和35.2%，分别超越Qwen3.5-9B基线2.4、8.7和2.9个百分点，达到基于环境的后训练水平。
原文摘要 · Abstract (English)
Program verifiers play a central role in training coding agents, including selecting trajectories for supervised fine-tuning (SFT) and providing rewards for reinforcement learning (RL). Standard execution-based verification requires running unit tests inside per-repository environments such as Docker images, incurring substantial environment setup costs. We propose Dockerless, an environment-free agentic patch verifier that evaluates generated code patches without executing them. Rather than simply matching candidate patches to references, Dockerless judges patch correctness using evidence gathered through agentic repository exploration. On a verifier evaluation benchmark, Dockerless outperforms the strongest open-source verifier by 14.3 AUC points. Using Dockerless as both the SFT trajectory filter and the RL reward enables a fully environment-free post-training pipeline. The resulting model reaches 62.0%, 50.0%, and 35.2% resolve rate on SWE-bench Verified, Multilingual, and Pro, respectively. It surpasses the Qwen3.5-9B baseline by 2.4, 8.7, and 2.9 points, matching environment-based post-training.
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。