Regularized self-improvement boosts AI agent performance and efficiency
RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
Machine LearningArtificial IntelligenceComputation and Language
Summary
Large language models (LLMs) work better when paired with smart systems that guide and manage them, called agent harnesses. The authors studied how these harnesses can improve themselves over time without just memorizing specific training tasks. They created a method called RRSI that carefully limits changes to keep improvements general and reusable. Their approach helped AI agents perform better on a variety of tasks while using less computational effort.
What this means in practice
- •For ai system developers: Improve the general efficiency and adaptability of AI agents across diverse tasks by applying RRSI to optimize their control systems and memory management.
- •For software engineering teams: Reduce the computational costs and improve performance of coding assistants and design automation tools by integrating RRSI-enhanced agent harnesses.
Authors
Peng Xia, Rujun Han, Zifeng Wang, Yanfei Chen, Yufan Zhang, Yoonho Lee, Chengsong Huang, Han Yu, Zhongying CuiZhu, Yifei Ming, Huaxiu Yao, Burak Gokturk, Tomas Pfister, Chen-Yu Lee
Abstract
An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Recent methods increasingly automate this process by iteratively proposing and selecting component-wise edits of an agent harness, practically establishing a form of recursive self-improvement (RSI) at the agent-system level. However, such recursive evolution may overfit by memorizing the training tasks, showing large in-distribution gains that shrink or even vanish on out-of-distribution benchmarks. We introduce Regularized Recursive Self-Improvement of Agent Harnesses (RRSI), which incorporates the principles of regularizations into harness self-improvement by constraining the evolution candidate proposal and selection. The proposer operates with a temporally annealed budget, limiting how many edits a candidate can bundle, and it encourages unexplored trajectories based on evolution history. The selector is equipped with a critic and a pruner: the critic screens benchmark-specific proposals, while the pruner, removes changes that are too small, too expensive, or no longer useful. Together these constraints favor reusable agent mechanisms over benchmark-specific ones or even noises. Across eight benchmarks spanning coding, agentic workspace and engineering design tasks, RRSI gains up to 14.1 points on the split it evolves against and up to 4.7 points on the five out-of-distribution benchmarks, while producing a harness that runs on 30% fewer policy tokens than the unregularized evolution. Code is available at https://github.com/google-research/rrsi and project page is https://regularized-rsi.com/.