We tackle the challenge of training reliable code-fixing agents in real repositories, where complex builds and shifting dependencies make evaluation unstable. We developed a verifiable pipeline with success defined as post-fix build validation and improved reproducibility across ~1K real issues by pinning dependencies and disabling automatic upgrades. Building on this, we introduced a scalable simplified pipeline for large-scale reinforcement learning (RL). Using this setup, we supervised fine-tuned Qwen3-32B in the full pipeline and applied RL on top of the SFT model in the simplified environment. The SFT model distilled from GPT-4.1 trajectories performs on par while being 56x smaller, and RL added 7-20% absolute gains under matched train-test conditions. "Thinking mode" was on par or worse in our experiments. Both SFT and RL models failed to generalize across environments, highlighting the importance of matching train-test environments for building reliable real-world code-fixing agents.
@article{arxiv.2510.22075,
title = {Agentic Reinforcement Learning for Real-World Code Repair},
author = {Siyu Zhu and Anastasiya Karpovich and Albert Chen and Jessica Koscheka and Shailesh Jannu and Di Wen and Yuqing Zhu and Rohit Jain and Alborz Geramifard},
journal= {arXiv preprint arXiv:2510.22075},
year = {2025}
}