Reproduce “ModHarness-RL: Recasting Harnesses as Learnable and Evolving Programs for Agentic RL” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zelin He, Siyu Zhu, Jacob Helwig, Yujian Liu, Li An, Rohit Jain, Shiyu Chang, Alborz Geramifard
ModHarness-RL makes an agent’s harness learnable through programs that reformat interaction history and provide guidance for recognizable failures. The authors fit compression before reinforcement learning and hold it fixed, while steering updates alternate with policy training. On CodeSearch, Qwen3-8B reaches 0.412 Overall F1 versus 0.333 for a fixed harness under limited context; this is the paper’s strongest full-system comparison.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.