Reproduce “RSI-Index: Measuring Autonomous LLM R&D in Frontier Agent Systems” 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
Vals AI Team
RSI-Index evaluates AI agents on four bounded tasks related to LLM development, independently scoring submitted artifacts against baselines and references. Its index averages task scores, with 0.5 marking reference parity. Among 19 systems, Claude Opus 5.5 leads at 0.373 and beats the reference only on LM Training; frontier-scale generalization remains untested.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.