Search papers, labs, and topics across Lattice.
This paper introduces Dockerless, an innovative environment-free program verifier designed for coding agents, which eliminates the need for costly Docker environments by assessing code patches through agentic repository exploration rather than execution. The method significantly improves verification efficiency, achieving a 14.3 AUC point increase over the leading open-source verifier on a benchmark evaluation. By integrating Dockerless into both supervised fine-tuning and reinforcement learning reward mechanisms, the approach enables a fully environment-free post-training pipeline, yielding competitive resolve rates compared to traditional methods.
Dockerless achieves a 14.3 AUC point improvement in program verification without the overhead of Docker environments, revolutionizing efficiency in training coding agents.
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.
