Reproduce “One Policy for 100 Household Tasks: Our Solution for the 2026 BEHAVIOR Challenge” 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
Ilia Larchenko, Gleb Zarin, Atindra Nair, Keerti Manney, Vitaly Kleban, Mirko Kiefer, Anjana Yodaiken
The authors train one warm-started policy across 100 household tasks and use a language-model agent that practices beforehand on 14 selected tasks where the policy struggles. The policy scores 0.392 Q across all tasks; the combined submission scores 0.446 Q and 27.3% full-task success. The agent’s advantage is limited to selected tasks, and the year-to-year gain has confounded causes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.