Reproduce “RSI-Env: An Open-Ended Environment for Recursive Self-Improvement of Agents” 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

Wanying Qu, Zhiyuan Fan, Jiyao Liu, Jiang Zhou, Jianghan Shen, Yanwei Fu, Dongrui Liu

RSI-Env provides a persistent workspace where agents can improve a model, its harness, or both, preserve research records, and let accepted systems continue later rounds. In a Knowledge Self run, SimpleQA accuracy rises from 82.0% to 85.0% to 88.0% across three versions. This demonstrates continuation in that run, not unlimited improvement; GPQA Diamond falls from 65.2% to 53.0%.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.