Reproduce “Intelligent Go-Explore: Standing on the Shoulders of Giant Foundation Models” 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
Cong Lu, Shengran Hu, Jeff Clune
Intelligent Go-Explore (IGE) extends Go-Explore by using foundation-model judgments to select archived states, choose actions, and filter newly reached states, while tracking tried actions for each state. On 100 hard Game of 24 problems, it solved all within 150 environment operations and was 70.8% faster than DFS; in Coin Collector, it achieved 88% versus 0% for others. Results concern deterministic benchmarks with simulator-based state restoration.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.